Techniques for insect detection in stored food grains: An overview

Accepted Manuscript Techniques for Insect Detection in Stored Food Grains: An Overview

Km.Sheetal Banga, Nachiket Kotwaliwale, Debabandya Mohapatra, Saroj Kumar Giri PII:

S0956-7135(18)30337-2

DOI:

10.1016/j.foodcont.2018.07.008

Reference:

JFCO 6221

To appear in:

Food Control

Received Date:

07 November 2017

Accepted Date:

06 July 2018

Please cite this article as: Km.Sheetal Banga, Nachiket Kotwaliwale, Debabandya Mohapatra, Saroj Kumar Giri, Techniques for Insect Detection in Stored Food Grains: An Overview, Food Control (2018), doi: 10.1016/j.foodcont.2018.07.008

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Techniques for Insect Detection in Stored Food Grains: An Overview

1

Km. Sheetal Banga1, Nachiket Kotwaliwale, Debabandya Mohapatra and Saroj Kumar Giri

2

ICAR-Central Institute of Agricultural Engineering, Bhopal, India-462038

3 4

Abstract

5

Insects cause a major loss in stored food grains. Besides, pestilential activities of insects

6

in stored food grains affect the marketability as well as the nutritional values. Early detection and

7

monitoring of insects in the stored food grains become necessary for applying corrective actions.

8

Visual inspection, probe sampling, insect trap, Berlese funnel, visual lures, pheromone devices

9

etc., are some of the popular methods largely used in commercial granaries or grain storage

10

establishments. Of late, electronic nose, solid phase micro-extraction, thermal imaging, acoustic

11

detection, etc. have been reported to be successful in detecting insects. The capability of in-situ

12

early detection, monitoring, cost, reliability, and labor requirements are the major factors

13

considered during for selection of the method. Detection of hidden infestation, whose population

14

may be many times higher than the free-living insects is an important concern to mitigate the

15

losses in bulk storage warehouses, so as to enable the early actions for fumigation or to dispose

16

off the grain. This paper reviews some of the widely used detection methods for early detection

17

of insects’ pestilential activities in stored food grains as well as some of the novel technologies

18

with an emphasis on acoustic method, which has a good commercial potential.

19 20

Highlights

1

Corresponding author

Email address: [email protected] ( Km. S. Banga)

1

ACCEPTED MANUSCRIPT

21



Traditional and modern methods for infestation detection in stored grains

22



Sensor technologies provide more reliable, automated and non-destructive detection

23



Acoustic sensor technology for early detection of infestation in stored grains

24

Keywords: Detection method, post-harvest losses, pestilential, acoustic detection, electronic

25

nose.

26 27

1.

Introduction

28

Global food security is an important issue as the world’s population is increasing rapidly

29

and will reach over 9.1 billion by the year 2050 (Parfitt et al., 2010 and FAO, 2014). About 20-

30

40% post-harvest losses (PHL) occur during field and post-harvest operations, and among these

31

losses 55% losses occur during storage (World Bank, 2011). The worldwide damage of food

32

grains due to insect infestation is estimated to be 10-40% annually (Asrar et al., 2016). In India,

33

storage losses for cereals were about 0.75-1.21%, while the losses in the case of pulses and

34

oilseeds were in the range of 1.18-1.67% and 0.22-1.61%, respectively (Jha et al., 2015). The

35

economic value of harvest and post-harvest losses of agricultural produces are estimated to be

36

INR 926510 million (based on production data of 2012-13 and wholesale prices of 2014, India)

37

(Jha et al., 2015). Ensuring availability of food to the ever increasing population will be a major

38

challenge in future due to a gradual decrease in the land and natural resources. It can be met

39

through the efficient use of agricultural produce and reducing the pre and post-harvest losses.

40

Sharon et al. (2014) reported that infestation caused the total quality loss as evident from the

41

reduction in seed germination, pH level and protein content, elevated moisture content and free

42

fatty acid content. Current standards of international trade apply only for external insects, and for 2

ACCEPTED MANUSCRIPT

43

the practical purpose, it is used as insect-free or insect absence checked by using standard

44

method (ISO 6322-3, 2001).

45

In Asia, about 6% of the total PHL occur due to improper storage facilities, in which

46

insects and fungi are responsible for the half (3%) of the storage losses (Sharon et al., 2014).

47

Specifically, whole pulses are at greatest risk, followed by the oilseeds and then by the cereal

48

grains for infestation (Sarwar, 2013). Food grains respire during storage; hence deterioration

49

occurs during storage, either quantitatively or qualitatively or both. Insects multiply at a higher

50

rate under favourable conditions and may even destroy 100 % of the grains and contaminate the

51

grains with their excreta and dead body parts resulting in with abominable odors and flavors

52

(Neethirajan et al., 2007). Insect infestation increases with the increase in temperature up to

53

45°C and high relative humidity due to their metabolic exertion, which aids to the growth of

54

microflora and the development of hotspots in stored food grains. At times, the whole lot of grain

55

can be charred due to increase in temperature, while stored in silo (Neethirajan et al., 2007).

56

Figure-1 indicates the losses occurred in various crops during pre and post-harvest operations,

57

from where it can be observed that losses due to insects are about 30% of the total loss.

58

Several detection techniques have been developed for the internal and external detection

59

of insects in stored food grains such as detection probe, staining of the kernel, Berlese funnel

60

method, acoustic techniques, uric-acid method, X-ray imaging, nuclear magnetic resonance

61

imaging, thermal imaging and solid-phase micro-extraction method (Neethirajan et al., 2007).

62

Some of these techniques are time-consuming, expensive, have potential health hazard, and less

63

efficient. Manual sampling traps and probes are the most common methods used on farms, while

64

manual inspection, sieving, and Berlese funnel method are used in grain storage and handling

65

facilities (Neethirajan et al., 2007). However, these methods are slow and are not able to detect 3

ACCEPTED MANUSCRIPT

66

the hidden infestation at early stages of primary pest species (Curculionidae, Bostrichidae,

67

Gelechiidae, and Bruchidae), whose population are ten times more than the free-living insects

68

(Fleurat-lessard, 2006).

69

Monitoring of stored food grains is used to ascertain the trends in insect’s number,

70

insect’s development stages or infestation level in a period of time. It also furnishes the insect's

71

activity in respect to environmental conditions and determines the efficacy of insect pest

72

management actions. To reduce these losses and to ensure the safe storage for sustainable

73

agriculture production, there is a need to develop advanced insect infestation detection methods

74

with higher sensitivity. While looking at traditional methods, this review paper is to give a birds-

75

eye view of the recently developed technologies for insect detection in stored products. Special

76

emphasis has been given to the acoustic detection of insects in stored food grains.

77

2.

Conventional methods of insect detection

78

Several conventional methods are used in grain storage establishments of which visual

79

inspection, probe sampling, and insect trap method are popular. These methods are simple but

80

time-consuming, labour-intensive and subjective. Some of the popular techniques are discussed

81

in brief in the following sections.

82

2.1 Detection of insect presence

83

2.1.1 Visual inspection

84

Detection of insect infestation in stored food grains can be done through visual inspection.

85

It is a uniform, qualitative and subjective method, used as a standard method for comparison of

86

quantitative methods (Semple, FAO report, 1980). Presence of eggs, adult insects, and infested 4

ACCEPTED MANUSCRIPT

87

grains can be seen by the naked eye without drawing grain samples or looking for residual

88

infestation within the storage bags. Ministry of Agriculture, Fisheries and Food Inspector,

89

Britain, developed some notations for the use of sack, storage, and sampling inspection (Table

90

1).

91

2.1.2 Probe sampling and trap method

92

Probe sampling and sieving are the most widely used methods; however it is laborious

93

and time-consuming. In this method, grains are drawn (0.5-1 kg) by probes from the stored bin.

94

Sieves are used for screening the insects from the grains. Probes are kept in grain storage bins for

95

long periods; an inspector manually removes it and visually inspects them, thus making it a time-

96

consuming and sometimes difficult procedure.

97

Various types of traps have been developed by TNAU, Coimbatore (India) (Mohan et al.,

98

1994). These devices (Figure 2) are useful in timely detection and monitoring of insect

99

infestation in stored food grains. Wandering of insects towards air is used as a concept to design.

100

The efficiency of two-in-one probe trap is high due to a combination of probe and pitfall trap. It

101

is best suited for trapping of pulse beetles as they always wander on the grain surface. An

102

indicator device consists of a perforated cone-shaped cup with a lid at the top, is fixed at the

103

bottom with a container and circular dish coated with a sticky material (Mohan, 2007; Mohan

104

and Rajesh, 2016). TNAU automatic insect removal bin removes the insect and crushes the eggs

105

laid by them. Its efficiency is very high (90%) as most of the insects can be removed within 10

106

days. It produced 1-4% damage in grains as compared to 33-65% damage in the ordinary bin

107

after 10 months storage period in paddy and sorghum grains (Mohan, 2007). The UV light traps

108

embedded with an ultra-violet light (4 W germicidal lamp) rays of 250 nm are used in storage 5

ACCEPTED MANUSCRIPT

109

godowns at 1.5 m above the ground level. These traps are useful in the trapping of a variety of

110

stored grain insects including lesser grain borer (Rhyzopertha dominica F.), rice weevil

111

(Sitophilus oryzae L.), red flour beetle (Tribolium castaneum Herbst), sawtoothed grain beetle

112

(Oryzaephilus surnamensis) etc. (Mohan and Rajesh, 2016).

113

2.1.3 Visual lures and Pheromones

114

“Light” can be used for detection, monitoring and management of insects in stored food

115

grains in warehouses, godowns, elevators etc. by utilizing the responses of insects towards the

116

light. It is a “clean” form of technology and uses three types of lights: incandescent, fluorescent,

117

and ultraviolet. Insects allured towards the lights of wavelength between 280-600 nm and some

118

colored objects due to their explicit reflectance (Neethirajan et al., 2007). Insect species, age,

119

environmental conditions, sex and intensity of light affects the responses of insects towards the

120

light (Shimoda and Honda, 2013). Pheromones are chemical substances secreted by the insects,

121

used in traps to control insect populations. These are used for communication among insects.

122

Traps of different materials, containing pheromones (sex and aggregation pheromone) used on

123

adhesive-coated surface or a funnel-shaped structure to catch the insects (Laopongsit and

124

Srzednicki, 2010).

125

2.2 Detection of insect density

126

2.2.1 Berlese funnel method

127

It consists of a standard Berlese funnel apparatus with a mesh screen (Minkevich et al.,

128

2002). Grain samples are put in the funnel below the incandescent light for 8 h and a jar

129

containing alcohol/water is used for capturing the insects. Funnels are equipped with screen

6

ACCEPTED MANUSCRIPT

130

bottom small enough to retain the grains and large enough to allow passage of the insects

131

through it. It uses dry heat to remove the insects from the grains. Dry heat warms the grains and

132

compels the insects to move opposite to heat in a funnel (Neethirajan et al., 2007).

133

2.2.2 Uric acid method

134

Uric acid, the main element of insect's excreta has been recommended as a tracing

135

element of insect infestation in stored foodgrains. This method detects the insect infestation of

136

the entire storage period indirectly (Rajendran, 2005). Later on, different methods were

137

developed to determine the uric acid level: by paper chromatographic, fluorometric, colorimetric,

138

gas-liquid chromatography (GLC), thin layer chromatography (TLC), high performance liquid

139

chromatography (HPLC) and by enzymatic methods (Ghaedian, 1995). As per BIS (1970),

140

colorimetric method can be used for uric acid measurement to determine the level of infestation.

141

2.2.3 Hidden infestation detector

142

It is a very simple and low-cost device used to detect hidden infestation in grains. It

143

consists of three circular plates placed over one another. The top and middle plates are hinged for

144

easy operation during lifting. The base plate is covered with ninhydrin treated filter paper.

145

Sorghum infested with S. oryzae, wheat with angoumois grain moth (Sitotroga Cerealella) and

146

green gram with cowpea weevil (Callosobruchus maculatus) were tested with this detector.

147

Holes of middle plate were filled with grain samples of about 20% moisture content. The top

148

plate was pressed to crush the grains. Filter paper stained the infested grains, which were counted

149

and percentage of infestation was estimated by comparing with other methods (Dakshinmurthy

150

and Ali, 1984; Anonymous, 1991).

151

3.

Modern methods of insect detection 7

ACCEPTED MANUSCRIPT

152

The approach of modern methods in stored food grains may offer an easy, rapid solution

153

to detect both internal and external infestation even of low density, through less destruction of

154

materials, so that decisive action can be taken as early as possible. Some of the technologies use

155

sensors, cameras, microscope, radiation sources, volatiles, sound etc. as measures for insect

156

detection. These methods need comparatively less labour than the conventional methods;

157

however, the labor should be skilled enough to control the sophisticated equipment as per the

158

protocols. These technologies can be grouped based on the properties employed for detection of

159

insects viz., electrical conductivity, olfactory, response to electromagnetic-spectrum and acoustic

160

signals. Details of the attempts made under these different categories are given in following

161

sections.

162

3.1 Conductance based method

163

3.1.1 Electrically conductive roller mill

164

In this method principle of electrical conductance and compression force is used for

165

infestation in stored foodgrains. In single kernel characterization system of two resistors and

166

voltage–divider circuit, one kernel acts as a resistor. The conductance of kernels is inspected

167

through the voltage during the crushing of kernels between the rolls. Presence of insects inside

168

the kernel increases the kernel moisture content, which provides an easy discrimination of sound

169

kernels from the infested kernel. This method is not suitable for detecting the insect eggs,

170

immature larvae and dead insects in low moisture grains (Pearson and Brabec, 2007).

171

Pearson et al. (2003) developed an electrically conductive roller mill "insect-o-graph" for

172

wheat classification. Infested kernels were classified from the uninfested kernels on the basis of

173

the signal characteristics of the system and the conductance signal received. Pearson and Brabec 8

ACCEPTED MANUSCRIPT

174

(2007) reported the detection of infested kernels above 70% along with larvae and pupae of R.

175

dominica, and tested 1 kg of wheat in about 2 min. Rice weevil (S. oryzae L.), and the lesser

176

grain borer (R. dominica F.), major internally infesting insects were tested. It can be a useful tool

177

for classification of grains at the receiving stations as it provides the information to grain storage

178

managers to make a decision whether a storage bin should be fumigated or to be discarded.

179

Brabec et al. (2010) investigated the detection of lesser grain borer (LGB) fragments in wheat

180

flour by using conductive roller mill. They found that conductance mill was suitable for testing

181

of internally infested insects in grains of low density (up to 3 insects) in 1-2 kg of grains at the

182

rate of 1 kg of grain per minute.

183

Brabec et al. (2012) used the modified laboratory mill to detect the internal infestation of

184

immature LGB in brown rice and wheat. The modified conductance mill had detected LGB in

185

500 g sample of brown rice and wheat by 97, 83, and 42% of large, medium, and small LGB

186

larvae, respectively within 150 s. The detection rate was higher for wheat as compared to thinner

187

brown rice. Brabec et al. (2017) detected the maize weevil (Sitophilus zeamais) infestation with

188

different development stages in popcorn kernels through a conductive roller mill. Two mills were

189

tested and it was observed that slower feeding mill could detect 81% pupae, 91% medium larvae,

190

and 47% small larvae while the faster mill detected 75% pupae, 80% medium larvae, and 43%

191

small larvae. Results show that this method is suitable for quick detection of matured stages of

192

larvae and pupae.

193

3.2 Olfactory based methods

9

ACCEPTED MANUSCRIPT

194

3.2.1 Solid phase micro-extraction (SPME)

195

Odour detection techniques for insect infestation and grain quality evaluation are gaining

196

popularity. In addition, this method facilitates early detection of infestation, storage age

197

determination, varietal discrimination of foodgrains etc. SPME used the headspace techniques to

198

isolate volatile compounds vaporized from samples, which was then condensed and finally

199

evaluated by gas chromatography-mass spectrometry (GCMS) for quantification of volatiles.

200

The efficiency and sensitivity of SPME method depend on extraction time and temperature.

201

High temperature and long extraction time favor in the collection of more analytes (Laopongsit

202

et al., 2014). Senthilkumar (2010) detected the T. castaneum and C. ferrugineus by headspace

203

analysis (HS-SPME) coupled with GCMS. Niu et al. (2016) used SPME coupled with GC-FID

204

(gas chromatography-flame ionization detection) and GC-MS to establish relationships between

205

storage period and grain quality, and grain quality and insect infestation of R. dominica in wheat.

206

Abuelnnor et al. (2010) identified distinct volatile compounds from infested wheat flour and

207

wheat grain with the T. confusum and S. granaries, respectively by SPME clenched with gas

208

chromatography-mass spectrometry. Larval and adult insects secreted distinct volatiles and these

209

distinct volatiles were useful for early monitoring of infestation.

210

3.2.2 Electronic nose (E-nose)

211

The intervention of divergent electronic nose (E-nose) sensor types and instruments,

212

works on the principle of electronic aroma detection (EAD) (Wilson, 2012). E-nose consists of

213

three parts: an odor sensor(s) set, a data pre-processor, and a data interpretation system. Sensor

214

set detects the volatile compounds present in the headspace of stored food grains and reacts by

215

changing the electrical properties. It is embedded with a predefined database to differentiate

216

certain volatiles (Wu et al., 2013). E-nose has the potential of rapid and automatic detection of 10

ACCEPTED MANUSCRIPT

217

insects in stored grains (Zhang et al., 2007). Selection of sensors array for specific volatile

218

organic compounds (VOCs) requires special attention to achieve the targeted results. Sensors

219

array should be selected to maximize the overall performance of the instrument and furnish the

220

different selectivity profiles for a particular application (Phaisangittisagu et al., 2010). Metal

221

oxide semiconductor (MOS) and conducting polymers (CP) type sensors are most widely used

222

but, carbon black composite (CBC), carbon dioxide (CO2), metal oxide semiconductor field

223

effect transistor (MOSFET), surface acoustic wave (SAW), optical fiber live cell (OF-LC) etc.

224

have also been used (Wilson, 2013).

225

Evans et al. (2000) used e-nose used to discriminate between infested and non-infested

226

samples of different fungal species through the development of secondary volatile metabolites.)

227

used the E-noses have been employed to discriminate and detect the insect infestation,

228

differentiate between insect species, predict insect population (Stetter et al., 1993; Wu et al.

229

2013) with some successes. Zhang and Wang (2007) used the E-nose to assess the detection of

230

storage age and insect (R. dominica F.) damage incurred in wheat.

231

3.3 Electromagnetic-spectrum based methods

232

3.3.1 Imaging methods

233

3.3.1.1 Machine vision within visible domain

234

Machine vision also known as computer vision, is an emerging technology which

235

combines the mechanics, optical instrumentation, electromagnetic sensing, digital and image

236

processing technology (Patel et al., 2012). It uses the principle of object recognition and

237

classification on the basis of information extracted from the image captured by using the camera 11

ACCEPTED MANUSCRIPT

238

(Sun, 2016). It is a rapid, consistent, economic and objective inspection technique that has the

239

potential for applications in quality evaluation of agricultural produce. The speed and accuracy

240

of this non-destructive technique can satisfy the demand of ever-increasing production and

241

quality requirements hence is helpful in intensifying the development of automated processes.

242

Machine vision technology consists of three main processes- image acquisition, image

243

processing or analysis and, recognition and interpretation. Image acquisition consists of

244

capturing a real image by using cameras, scanners, videos etc. and transforming it into a digital

245

image (Figure 3). Image pre-processing assigns the initial processing of the raw image.

246

Sometimes, pre-processing is accomplished to enhance the image quality by suppressing

247

undesired distortions “noise” or by the advancement of important features of view of interest

248

(Narendra and Hareesh, 2010). Image feature extraction comprised the extraction of image

249

features at distinct levels of intricacy from the image data (Davies, 2005). In image segmentation

250

process, cutting, adding and feature analysis of images is done to divide image regions that have

251

a substantial correlation with objects or view of interest using the principle of matrix analysis. It

252

is one of the most crucial steps of the whole image processing technique, as the accuracy of this

253

technique is extremely dependent on consequent extracted data. Image recognition and

254

interpretation provide the useful information after image analysis that can be used for process or

255

machine control (Gunasekaran, 1996). Monitoring and evaluation of varieties and soundness of

256

grains, amount of foreign material, mold and insect infestation etc. in bulk grains can be

257

successfully carried out (Aviara et al., 2016). This technique is suitable to detect the whole and

258

live insects in stored food grains. Zayas and Flinn (1998) used the multispectral analysis with

259

pattern recognition to detect the R. dominica in bulk wheat samples and the results revealed that

260

recognition efficiency for adult lesser grain borer and some foreign material was more than 90%.

12

ACCEPTED MANUSCRIPT

261

3.3.1.2 X-ray imaging

262

X-ray imaging method is an encouraging technique utilized the non-contact sensor for

263

inspection of large samples while appreciably providing the information (Yacob et al., 2005).

264

Soft X-ray imaging is a fast non-destructive and direct method, used for the detection of invisible

265

insects in stored food grains (Karunakaran et al., 2003), grading, internal quality of agricultural

266

produce also for hidden insects in mangoes (Kotwaliwale et al., 2014). The imaging system

267

encompassed with X-ray source; X-ray converter; imaging system and isolated casing for image

268

capture (Kotwaliwale et al., 2011). The imaging medium, captured the images, isolated from the

269

surrounding radiation by a casing (Kotwaliwale et al., 2007). Electromagnetic waves of 0.1-

270

10 nm wavelengths with 0.12-12 KeV energy are used as soft X-ray for internal inspection in

271

very less time (3-5 s) to produce a X-ray image. Kotwaliwale et al. (2007a) determined the

272

quality of pecans (Carya illinoinensis (Wangenh.) K. Koch) by using soft X-ray with voltage

273

levels ranging from 15 to 50 kVp in steps of 5 kVp, and five current levels (0.1 to 1.0 mA) and

274

with two orientations. They found that severely insect damaged samples and severely

275

underdeveloped nutmeat could be isolated easily due to low percent area occupancy. Features of

276

pecan such as shell, nutmeat, air gap between shell and nutmeat, defects, and presence of insects

277

were distinctly visible in X-ray images after contrast stretching (Figure 4). Karunakaran et al.

278

(2003a) used the soft X-ray method in detecting Cryptolestes ferrugineus (Stephens), T.

279

castaneum (Herbst), Indianmeal moth (Plodia interpunctella), S. oryzae (L.), and R. dominica

280

(F.) in wheat kernels. The parametric classifier identified infested kernels by different stages of

281

S. oryzae and larvae of R. dominica with more than 98% accuracy.

282

Chelladurai et al. (2014) used the soft X-ray and near-infrared (NIR) hyperspectral

283

imaging techniques to acquire images of soybeans infested by C. maculatus along with 13

ACCEPTED MANUSCRIPT

284

uninfested kernels. 33 features were extracted by soft X-ray imaging and 48 features were

285

extracted by hyperspectral imaging for data analysis. Different stages of infestation were

286

classified by using LDA and quadratic discriminant analysis (QDA) models. About 86% of

287

uninfested grains and 83% of infested grains by C. maculatus were classified by LDA for soft X-

288

ray images. Hyperspectral data were classified by the principal component analysis (PCA) at

289

wavelengths of 960 nm, 1030 nm, and 1440 nm. Combination of X-ray and hyperspectral

290

features improved the classification accuracy of egg and larvae stages.

291

3.3.1.3 Thermal imaging

292

Thermal imaging is a night vision technology, which improves the visibility of an object

293

in a dark environment by detecting the infrared radiation energy emitted by the object and

294

transforming it into the visible image. Infrared energy emitted by an object is a function of

295

temperature and known as heat signature and the produced image is known as thermogram

296

(Nanje Gowda et al., 2013). This imaging system consists of detectors and lenses, thermal

297

imaging cameras and data collection tools. This technique is useful in the management of stored

298

food grains by detecting damaged grains, foreign materials, and internal and external infestation

299

of grains. It is useful, where temperature differences are used in evaluation or quantification of a

300

process. Thermal imaging provides the potential advantages over the fluorescence and

301

hyperspectral imaging in cost reduction and determination of material properties.

302

Detection of Rusty grain beetle (Ceyptolestes ferrugineus) infestation in the wheat kernel

303

by infrared thermal imaging with the camera, control panel, microwave applicator, conveyor etc.

304

has been reported by Manickavasagan et al. (2008). They used the system for detection of C.

305

ferrugineus inside the seed coat on the germ of wheat kernels. Thermal images were captured for

306

the infested kernels to identify the developmental stages of the insect. Respiration rate of 14

ACCEPTED MANUSCRIPT

307

different stages of C. ferrugineus were highly correlated(r = 0.83-0.91) with the temperature

308

distribution on the surface of the infested kernels. Classification accuracy of quadratic and linear

309

function for infested and sound kernels was 83.5% and 77.7% and, 77.6% and 83.0%,

310

respectively. Chelladurai et al. (2012) used the thermal imaging for moong bean infested by C.

311

maculatus. Classification accuracy of 55.24-77.84% by LDA model and 75.45-91% by QDA

312

model were obtained. More than 80% moong beans were identified as infested by initial stages

313

of C. maculatus with QDA model. Thermal images of fungal infected paddy were obtained using

314

mid infrared thermal camera at 0 s (before heating), 180s (after heating) and 210 s (30 s after

315

cooling) (Figure 5). Average pixel of the image was used as feature to determine the moisture

316

content. Fungal infected paddy gave higher average pixel values compared with non-fungal

317

paddy (Khairunniza-Bejo and Jamil, 2013).

318

3.3.2 Non-imaging methods

319

3.3.2.1 Electronic grain probe insect counter (EGPIC)

320

An automated passive grain probe, known as Electronic Grain Probe Insect Counter

321

acquires offsite monitoring and detection of insect pests and remotely displays the data of

322

infestation levels in stored food grains. It consists of a probe, system circuitry, data logger and a

323

user interface. Due to the danger of grain dust explosions, electric power and circuitry are kept

324

outside from the storage structure and pass only low voltage, high impedance, sensor leads

325

through the grains from the beam generation/detection circuitry to sensor head. Acquired

326

information by sensors is transmitted to the computer that analyses the signals and makes time-

327

stamped records of detection (Litzkow et al., 1997). When an insect falls on the sensor head of

328

EGPIC across the infrared beam, signal is generated by the infrared diode and is then received by 15

ACCEPTED MANUSCRIPT

329

the infrared phototransistor, which makes a slight declination in the light intensity passing to the

330

phototransistor. The electronic circuitry system detects this slight declination and converts it into

331

a time-stamped insect count and then insects exit the probe (Shuman and Epsky, 2001). It

332

monitors the insect’s levels continuously at any depth. Flinn et al. (2009) tested the commercial

333

electronic grain probe trap in two bins of wheat (32.6 tonnes) during 2 years of storage. They

334

developed a regression model and compared it with the insect density estimated by EGPIC.

335

There was 40-75% variation in predicted insect density against EGPIC. An expert system

336

“Stored Grain Advisor Pro” was coupled with EGPIC to estimate the C. ferrugineus (Stephens),

337

R. dominica (F.), and T. castaneum (Herbst) density from trap catch counts. Expert system

338

estimated insect density efficiently but it was unable to differentiate the R. dominica and T.

339

Castaneum due to the same size (Flinn et al., 2009).

340

3.3.2.2 NIR spectroscopy

341

Near-infrared spectroscopy (NIRS) measures the concentration of biological materials

342

such as water, protein, starch etc. by taking the dispersed reflectance, interactance or

343

transmittance of the sample in the range of 780-2500 nm. It is a non-destructive, fast, accurate

344

and cost-effective method viable for internal as well as external detections in fruits, vegetables,

345

cereals and pulses (Elizabeth, 2002; Kim et al., 2003; Xing and Guyer, 2008). Reflectance mode

346

measures the light reflected or dispersed back from the surface of the object. Interactance mode

347

is acclamatory when transmittance assessment is arduous to access the internal information

348

(Kavdir et al., 2007). Ridgway and Chambers (1996) reported the use of NIR reflectance

349

spectroscopy in the detection of internal infestation by Sitophilus granarius (L.) in wheat.

350

Chelladurai et al. (2014) used soft X-Ray imaging and NIR hyperspectral imaging in the

351

detection of C. maculatus and classified the infested and sound kernels of soybean. Maghirang et 16

ACCEPTED MANUSCRIPT

352

al. (2003) detected the live pupae and larvae of S. oryzae (L.) with an accuracy of 92% to 93%

353

by an automated NIRS system (400-1700 nm). A comparison was done by Perez-Mendoza et al.

354

(2003) between a NIRS and standard floatation method for wheat flour and they found that NIRS

355

was the rapid method (less than one min/sample), could be able to detect in bulk samples without

356

any sample preparation. Fourier transformation near-infrared (FT-NIR) technology, a rapid

357

detection instrument was used to detect moldy maize kernels with a 86.7% validation accuracy

358

on characteristic wavelengths of 1466 nm, 1530 nm, 1926 nm, 2321 nm and 2384 nm (Chu et al.,

359

2014).

360

3.4 Acoustic detection

361

Acoustic technology depends on the hypothesis that the sound made due to movement

362

and feeding of insects can be monitored to estimate the type and density of insects within a

363

stored grain mass. It has shown encouraging results on detection of internal and external insects

364

in the grain mass during early stages of infestation through insect feeding sounds (Eliopoulos et

365

al., 2015). Acoustic sensors detect the mechanical or acoustic waves. When an acoustic wave

366

transmits through certain material or object, it gets affected by the properties of material/object

367

and by any hurdle present in the path. Due to this, velocity or amplitude of the acoustic waves

368

dampens and then these changes are translated into digital or analog signal through transducers.

369

It generally uses the piezoelectric substrate as a sensor. Detection of concealed insects in grain

370

kernel depends on amplification and filtering of their movement and feeding sounds.

371

Classification of targeted sounds from other sounds and other limiting factors such as sensor

372

sensitivity, sound-noise ratio, the range of sensors etc., limit the applicability of acoustic devices.

373

With technological advancement, improved sensors and the use of digital signal processing 17

ACCEPTED MANUSCRIPT

374

software tools has enabled increased receptiveness and the credibility. Pattern features (spectral

375

and temporal) are also useful in classification of background noise from the targeted noise.

376

Standard speech recognition tools like Gaussian mixture models, hidden Markov models were

377

used for the separation of insects’ sounds from background noise (Mankin et al. 2009).

378

Attempts were also made to detect the damage and infestation in wheat and hazelnuts by

379

using impact acoustic technology and voice recognition methods (Pearson et al., 2007). Acoustic

380

detection of internal feeding and movement sounds of S. zeamais, rice weevil (S. oryzae (L.))

381

and granary weevil (S. granarius (L.)) were at relatively low intensity, 15-35 dB with frequency

382

2-6 kHz (Pittendrigh et al., 1997). Generally, the sound of insects comprises of trains of short

383

broadband impulses (1-10 ms), whereas background noise occurrs as continuous signals with

384

symphonic peaks (Mankin et al., 2011). Hagstrum et al. (1996) acoustically detected the R.

385

dominica (F.), T. castaneum (h.) and S. oryzae (L.) in wheat grains. The sound produced by

386

insects was detected without removing grain samples from the storage bin. Above 90% detection

387

level could be achieved at the bottom of the structure. They suggested it as an easy and quick

388

method of detection and population density estimation of insects. Kiobia et al. (2015) developed

389

a sound and vibration controlled system to detect Prostephanus truncates and S. Zeamais in

390

maize storage for managing the pest in stored food grains in Sub-Saharan Africa. They found

391

that larval impulses were measurable by the sensors within 25 cm range. Software tools were

392

also developed on the basis of differences in the spectral and temporal pattern of insect sounds

393

correlated with differences in physiological activities. Feeding sounds produced broad, high-

394

frequency spectrum than the low energy moving sounds (Mankin et al., 2010). Trains of

395

consecutive insect impulses (200 ms or less) established more significant indicator of insect

396

presence than the single impulse (Njoroge et al., 2016, 2017). 18

ACCEPTED MANUSCRIPT

397

Fleurat-Lessard et al. (2006) developed a fully automatic system for the inspection of

398

insects in bulk grains of wheat and a pilot scale system of 300 kg (Figure 6). Detection of

399

primary pests of adult and larvae of S. oryzae (SO) at a density of one adult/10kg and one

400

larva/kg of wheat within 2 min scanning time with more than 95% of likelihood could be

401

achieved by this system in 20 cm detection range with a sample of 65 kg. Detection of R.

402

dominica (RD) was easier than SO. For SO, peak sound was 3.3-3.8 kHz and for RD was 2.6 to

403

3.2 kHz. The relationship between insect activity and density levels was quantitatively modeled

404

in the range from one individual per 10 kg to 10 individuals per kg at temperature levels ranging

405

from 5 to 30 °C. Threshold temperature for larval activity was as low as 8°C (SO) and 15°C for

406

RD.

407

Lebnac et al. (2011) developed a real-time acoustic insect detection probe to detect and

408

identify the sounds of different species of different stages of primary insects in long-term

409

storage, operated from the surface of the bulk grain. The equipment possesses the prediction of

410

live concealed insects of all stages with a confidence factor greater than 90 % in a 30 kg grain

411

mass. Eliopoulos et al. (2015) reported the efficacy of bioacoustics in detecting the presence of

412

adult beetles inside the grain mass of wheat. Adults of the most important beetle pests of stored

413

cereals and pulses, in various population densities (1, 2, 10, 20, 50, 100, 200 & 500 beetle

414

adults/kg grain) were used. The linear model was very effective in describing the relationship

415

between population density and number of sounds. Njoroge et al. (2016) investigated the

416

frequency and time pattern differences in acoustic signals produced by Prostephanus truncates

417

(Horn) and S. zeamais (Motschulsky) in stored maize to distinguish among these species and

418

stages. Frequency profiles were categorized into five types of profiles to show the differences

419

between peak energy and broadness of frequency range. Bursts of three closely spaced impulses 19

ACCEPTED MANUSCRIPT

420

were observed and mean rates of bursts, mean counts of impulses per burst, and mean rate of

421

impulses in bursts were calculated to compare the species and stages.

422

Bioacoustics of bean weevil (Acanthoscelides obtectus) on Common bean (Phaseolus

423

vulgaris) was used to see its emergence inside the grains in an anechoic chamber. Low amplitude

424

sound impulses at low and higher rate were used in real-time detection and classification of

425

insect sound from background noise (Njoroge et al., 2017). Advancement in research of such

426

technology in the future will reduce its cost development of acoustic detection mobile app with

427

existing data for specific species for farmers use.

428

3.4.1 Signal processing

429

Processing of recorded sound for classification of background noise and targeted noise uses

430

speech recognition methods. For the analysis of acoustic signals, various types of windows are

431

used for spectrogram analysis.

432

3.4.1.1 Window Function and filtering

433

In signal processing, a window function (also known as an apodization or tapering

434

function) is a mathematical function, used to smoothly draw a sampled signal down to zero at the

435

edges of the sampled region (Prabhu, 2013). The spectrum of window function can be given in

436

time-domain as well as in the frequency domain. In the analysis of acoustic signals, more than

437

one sample (atleast two full cycles) is used. In the window, a series of speech samples are

438

selected for spectral analysis. Different types of windows are used for acoustic detection system,

439

viz. rectangular, bartlett, hamming, hanning, blackman, blakman-harris, welch and gaussian

440

(Prabhu, 2013). Rectangular window is not used for frequency analysis of the acoustic signal

441

(Mannell, 2008). Hanning window uses cosine cycles, valued between 0 and 1. Hamming

442

window is related to hanning and it covers the values between 0.054 and 1. Generally, gaussian 20

ACCEPTED MANUSCRIPT

443

windows are used for the analysis of acoustic signals (Mannell, 2008) in biological materials.

444

Complex acoustic signals are filtered prior to processing via low-pass (LP) filter, high pass (HP)

445

filter and band pass (BP). During filtering, some frequencies are allowed to pass and others

446

should be blocked. In most filters, there is a region around the cut-off frequency where

447

frequencies are partially allowed to pass. This provides a more gentle transition between the

448

pass-band (the frequencies which are unattenuated) and the stop-band (the frequencies which are

449

attenuated) (Mannell, 2008).

450

The duration of R. ferrugineus sound impulse was 3-50 ms and 3.8 kHz peak frequency

451

range (Potamitis et al., 2009). Usually, the peak of relative energy of background noise was high

452

at low frequency (below 1 kHz) (Mankin, 2010). Requisite of filtering depends on the nature and

453

intensity of the background noise. After filtering, signals above the threshold amplitude were

454

removed. The amplitude and frequency of insects were affected by the installed locations of

455

sensors. If multiple sensors are used then signals from one sensor can be subtracted from the

456

other to reduce distant noise before further signal processing.

457

3.4.1.2 Acoustic spectrum features

458

Discrimination of signal-noise can be done by separating respective sound produced by the

459

insects by computing certain features of each impulse and then they are compared with the

460

normalized spectral features (Potamitis et al., 2009). Several spectral features were reported to be

461

useful for discrimination viz. Fourier transform, the dominant harmonic, and linear frequency

462

cespstral coefficients (Mankin et al., 2009). Clustering of features of insect produced sound have

463

been used to discriminate the background noise features viz. vector quantization and Gaussian

464

mixture models (Potamitis et al., 2009).

21

ACCEPTED MANUSCRIPT

465

3.4.1.3 Acoustic temporal pattern features

466

The authenticity of insect detection can be enhanced by consolidating temporal pattern

467

features. Many insects possess behavioral patterns with regularities. Some of the patterns with

468

insufficient regularities were not characterized reliably by the computer programs, but some

469

studies found that these patterns were identified as a burst of impulses separated by quiet

470

intervals of 0.25 s or more (Mankin et al., 2009). About six impulses occurred in a burst due to

471

feeding or movement of insects. This phenomenon can be used as signal features to discriminate

472

a target insect sound from the background sounds. Identification of temporal features is easy for

473

large and active insects. Consolidation of bursts as a signal processing feature enables the

474

removal of wind-induced trapping noise or other background noise similar to the sound signals

475

produced by insects. A detailed description of advantages and limitations of modern methods employed for

476 477

insect detection are tabulated in Table 2.

478

4.

Conclusions

479

Several methods are available to detect the insect infestation in stored foodgrains. Among

480

conventional methods, visual inspection is a simple, direct and inexpensive method but not

481

suitable for bulk storage, detection of hidden and low-density infestation; besides being time-

482

consuming. Sampling probes and traps can be effective but are time-consuming and tedious in

483

nature and provide only the temporal data and sometimes need destruction of samples. The

484

visual lure is a chemical-free method which attracts the insect by light but their accuracy is

485

affected by environmental factors and offers low sensitivity and gives information only about

486

adult insects. Pheromones method does not need sampling and it can detect the internal as well 22

ACCEPTED MANUSCRIPT

487

as external insects, however its detection effciency is limited to few species of insects as limited

488

information is available on different pheromone characteristics. Berlese funnel method is one of

489

the most common methods used in grain elevators, but it is slow, unable to detect hidden

490

infestation and accuracy depends on insects’ size. The uric acid detection method is an officially

491

accepted method in many developing countries, however it is applicable to highly infested grains

492

and produces low sensitivity. Hidden infestation detector is a simple, cost-effective apparatus but

493

it is applicable for a very small quantity, and destructive in nature. Most of the conventional

494

methods show effectiveness for external detection only.

495

In recently developed methods, electrical conductance method detects the hidden infestation but

496

it cannot detect the egg and larvae stages and requires high moisture content and applicable only

497

for single grain characterization at a time. The solid phase micro extraction method coupled with

498

dynamic headspace has high sensitivity but it can detect only the adult stages and requires the

499

skilled person for the operation and analysis. An electronic nose in the detection of insects

500

provides the quick, objective, less poisoning high technology but its limitation lies in the

501

requirement of a large amount of experimental data to train the sensors; besides the sensors are

502

affected by the environmental factors and also their efficiency decreases over time. Machine

503

Vision system operating in visible range is used for various applications in agricultural

504

operations and it shows the ability to classify the grains but it is expensive and unable to detect

505

the dead and internal insects. X-ray is a non-destructive and direct method, applicable for

506

internal as well as external detection but it is costly and harmful as it requires the skilled person.

507

Thermal imaging has the potential to identify the infested and sound grains but it cannot identify

508

the development stages of insects and it is low in cost as compared to the hyperspectral and

509

fluorescence imaging. EGPIC is a real-time automatic monitoring method suitable for bulk

23

ACCEPTED MANUSCRIPT

510

storage but it restricts the placement of EGPIC to easily accessible locations. NIR spectroscopy

511

cannot detect the low level of infestation and unable to differentiate between the live and dead

512

insects but it is very sensitive to the moisture content of samples. Acoustic detection technique

513

demonstrates the superior sensitivity in internal and external infestation as well as it is low in

514

cost, automatic, rapid, non-destructive and able to perform well in bulk storage of grains.

515

Discrimination of background noise from the insect’s sound spectrum is a challenging task. The

516

advent of various software, technology, and statistical tools reduced the complexity in the

517

analysis of capture sound spectrums. Therefore, it has a potential to be used in bulk storage of

518

food grains for early detection of insect infestation. Some researchers suggest large-scale

519

application of this technology to ensure the safety of stored agriculture produces. There is a need

520

for such a potential technology which caters automation, high efficiency in bulk storage, less

521

human power and little or no destruction of the commodity. With the advancement of

522

computational power and availability of very good sensing devices at affordable cost, acoustic

523

technology emerges as a promising technique to non-destructively detect and continuously

524

monitor insects in stored food grains.

525

References

526

1.

Abuelnnor, N., Jones, P. R. H., Ratcliffe, N. M., de Lacy Costello, B., & Spencer-Phillips,

527

P. T. (2010). Investigation of the semiochemicals of confused flour beetle

528

Triboliumconfusum Jaquelin du Val and grain weevil Sitophilus granarius (L.) in stored

529

wheat grain and flour. Integrated Pest Management Reviews, 7(2), 105-121.

530

2.

Anonymous (1991). AICRP on Post-Harvest Technology (1976-90). CIAE, Bhopal, pp-5356.

531 532

3.

Anonymous (2016). http://precisionce.com/berlese-funnel-extraction-unit

533

4.

Asrar, M., Ashraf, N., Hussain, S. M., Zia, K., & Rasool, B. (2016). Toxicity and

534

repellence of plant oils against Tribolium castaneum (herbst), Rhyzopertha dominica (f.)

535

and Trogoderma granarium (e.). 24

ACCEPTED MANUSCRIPT

536

5.

Aviara, N. A., Fabiyi, O. E., Ojediran, J. O., Ogunniyi, O. I., & Onatola, I. T. (2016).

537

Application of computer vision in food grain quality inspection, evaluation and control

538

during bulk storage. 37th Annual conference and annual general meeting–“minna 2016”.

539

6.

Physics.

540 541

7.

BIS (1970) IS: 4333 (part V) - Determination of uric acid. Bureau of Indian standards, New Delhi.

542 543

Beranek, L. L. (1988). Acoustical measurements. Woodbury, NY: American Institute of

8.

Brabec, D., Pearson, T., Flinn, P., & Katzke, D. (2010). Detection of internal insects in

544

wheat using a conductive roller mill and estimation of insect fragments in the resulting

545

flour. Journal of Stored Products Research, 46(3), 180-185.

546

9.

Brabec, D., Pearson, T., & Flinn, P. W. (2012). Detection of Lesser Grain Borer Larvae in

547

Internally Infested Kernels of Brown Rice and Wheat Using an Electrically Conductive

548

Roller Mill1. Applied Engineering in Agriculture, 19(6), 727-733.

549

10.

Brabec, D., Dowell, F., Campbell, J., & West, M. (2017). Detection of internally infested

550

popcorn using electrically conductive roller mills. Journal of Stored Products

551

Research, 70, 37-43.

552

11.

grain and their control. Advances in Cereal Science and Technology.

553 554

Bulla Jr, L. A., Kramer, K. J., & Speirs, R. D. (1978). Insects and microorganisms in stored

12.

Chelladurai, V., Kaliramesh, S.J., & Jayas, D.J., 2012. Detection of Callosobruchus

555

Maculatus (F.) Infestation in Mung Bean (Vignaradiata) using Thermal Imaging

556

Technique. NABEC/CSBE 12-121.

557

13.

Chelladurai, V., Karuppiah, K., Jayas, D. S., Fields, P. G., & White, N. D. G. (2014).

558

Detection of Callosobruchus maculatus (F.) infestation in soybean using soft X-ray and

559

NIR hyperspectral imaging techniques. Journal of Stored Products Research, 57, 43-48.

560

14.

Chu, X., Wang, W., Yuan, Y., Xi, M., Yoon, S. C., & Peng, Y. (2014, July). Detection of

561

Mildew Maize Kernels Using FT-NIR Spectroscopy. In 2014 Montreal, Quebec Canada

562

July 13–July 16, 2014 (p. 1). American Society of Agricultural and Biological Engineers.

563

15.

Dakshinmurthy, A. & Ali, N. (1984). A simple device for determining hidden infestation in

564

grains. Proc. Of the XXI Annual Convention of ISAE held at New Delhi during April,

565

1984. 25

ACCEPTED MANUSCRIPT

566

16.

Kaufmann, New York. ISBN 0-12-206093-8.

567 568

Davies, E. R. (2005). Machine Vision: Theory, Algorithms, Practicalities. Morgan

17.

Elizabeth, B. M., Dowell, F. E., Baker, J. E., & Throne, J. E. (2002). Detecting single

569

wheat kernels containing live or dead insects using near infrared reflectance spectroscopy.

570

ASAE Paper No. 023067. Chicago, IL: ASAE.

571

18.

Eliopoulos, P. A., Potamitis, I., Kontodimas, D. C., & Givropoulou, E. G. (2015).

572

Detection of adult beetles inside the stored wheat mass based on their acoustic emissions.

573

Journal of Economic Entomology, 108(6), 2808-2814.

574

19.

probe insect counter. Journal of stored products research, 37(2), 187-197.

575 576

20. 21.

FAO (2009). How to Feed the World in 2050. Rome: FAO. Available from: http://www.fao.org/fileadmin/templates

579 580

Food and Agriculture Organization of United Nations (2014). Global Initiative on Food Losses and Waste Reduction; FAO: Rome, Italy.

577 578

Epsky, N. D., & Shuman, D. (2001). Laboratory evaluation of an improved electronic grain

22.

Evans, P., Persaud, K. C., McNeish, A. S., Sneath, R. W., Hobson, N., & Magan, N.

581

(2000). Evaluation of a radial basis function neural network for the determination of wheat

582

quality from electronic nose data. Sensors and Actuators B: Chemical, 69(3), 348-358.

583

23.

insect pest infestation risks during cereal storage. Industries des Céréales, 148.

584 585

Fleurat-Lessard, F. (2006).Advanced methods for the monitoring of fungal spoilage and

24.

Fleurat-Lessard, F., Tomasini, B., Kostine, L., & Fuzeau, B. (2006). Acoustic detection

586

and automatic identification of insect stages activity in grain bulks by noise spectra

587

processing through classification algorithms. In Proc. 9th International Working

588

Conference on Stored Product Protection (pp. 15-18).

589

25.

Flinn, P. W., Opit, G. P., & Throne, J. E. (2009). Predicting stored grain insect population

590

densities using an electronic probe trap. Journal of Economic Entomology, 102(4), 1696-

591

1704.

592

26.

methods of analysis.

593 594 595

Ghaedian, A. R. (1995). Detection of insect infestation in stored products by instrumental

27.

Gunasekaran, S. (1996). Computer Vision Technology for Food Quality Assurance review. Trends in Food Science and Technology, 7: 245-256.

26

ACCEPTED MANUSCRIPT

596

28.

Hagstrum, D. W., Flinn, P. W., & Shuman, D. (1996). Automated monitoring using

597

acoustical sensors for insects in farm-stored wheat. Journal of Economic Entomology,

598

89(1), 211–217.

599

29.

infestation, Part 3: Reference Method.

600 601

ISO 6639-3 (2001) (English): Cereals and Pulses - Determination of hidden insect

30.

Jha, S. N., Vishwakarma, R. K., Ahmad, T., Rai, A., & Dixit, A. K. (2015). Report on

602

assessment of quantative harvest and post-harvest losses of major crops and commodities

603

in India. ICAR-All India Coordinated Research Project on Post-Harvest Technology,

604

ICAR-CIPHET, PO-PAU, Ludhiana-141004.

605

31.

kernels infested by Sitophilus oryzae. Transactions of the ASAE, 46(3), 739–745.

606 607

32.

Karunakaran, C., Jayas, D. S., & White, N. D. G. (2003a). X-ray image analysis to detect infestations caused by insects in grain. Cereal Chemistry, 80(5), 553-557.

608 609

Karunakaran, C., Jayas, D. S., & White, N. D. G. (2003). Soft X-ray inspection of wheat

33.

Kavdir, I., Lu, R., Ariana, D., & Ngouajio, M. (2007). Visible and near-infrared

610

spectroscopy for nondestructive quality assessment of pickling cucumbers. Postharvest

611

Biology and Technology, 44(2), 165-174.

612

34.

Khairunniza-Bejo, S., & Jamil, N. (2013). Preliminary Study on Detection of Fungal

613

Infection in Stored Paddy Using Thermal Image. International Proceedings of Chemical,

614

Biological and Environmental Engineering, 60, 19-23.

615

35.

infrared reflectance spectroscopy. Cereal Chemistry, 80(3), 346–349.

616 617

Kim, S. S., Phyu, M. R., Kim, J. M., & Lee, S. H. (2003). Authentication ofrice using near

36.

Kim MG, Yang JY, & Lee HS (2013). Phototactic behavior: repellent effects of cigarette

618

beetle, Lasiodermaserricorne (Coleoptera: Anobiidae), to light-emitting diodes. Appl Biol

619

Chem 56:331–333.

620

37.

Kiobia, D. O., Tumbo, S. D., Cantillo, J., Rohde, B. B., Mallikarjunan, P. K., & Mankin,

621

R. W. (2015). Characterization of sounds in maize produced by internally feeding insects:

622

investigations to develop inexpensive devices for detection of Prostephanustruncatus

623

(Coleoptera: Bostrichidae) and Sitophilus zeamais (Coleoptera: Curculionidae) in small-

624

scale storage facilities in sub-Saharan Africa. Florida Entomologist, 98(2), 405-409.

625 626

38.

Kumar, D., & Kalita, P. (2017). Reducing Postharvest Losses during Storage of Grain Crops to Strengthen Food Security in Developing Countries. Foods, 6(1), 8. 27

ACCEPTED MANUSCRIPT

627

39.

Kotwaliwale, N., Kalne, A. & Singh, K. (2011).Radiography, CT and MRI. In: Jha SN,

628

editor. Book chapter in non-destructive evaluation of food quality-theory and

629

practice. Berlin: Springer.

630

40.

Kotwaliwale, N., Singh, K., Kalne, A., Jha, S. N., Seth, N., & Kar, A. (2014). X-ray

631

imaging methods for internal quality evaluation of agricultural produce. Journal of Food

632

Science and Technology, 51(1), 1-15.

633

41.

Kotwaliwale N, Subbiah J, Weckler PR, Brusewitz GH, & Kranzler GA. (2007).

634

Calibration of a soft X-ray digital imaging system for biological materials. Trans

635

ASABE. 50(2):661–666.

636

42.

Kotwaliwale, N., Weckler, P. R., Brusewitz, G. H., Kranzler, G. A., & Maness, N. O.

637

(2007a). Non-destructive quality determination of pecans using soft X-rays. Postharvest

638

Biology and Technology, 45(3), 372-380.

639

43.

grain based on head space analysis of volatile compounds. Julius-Kühn-Archiv, (425), 999.

640 641

Laopongsit, W., & Srzednicki, G. (2010). Early detection of insect infestation in stored

44.

Laopongsit, W., Srzednicki, G., & Craske, J. (2014). Preliminary study of solid phase

642

micro-extraction (SPME) as a method for detecting insect infestation in wheat

643

grain. Journal of Stored Products Research, 59, 88-95.

644

45.

Leblanc, M. P., Gaunt, D., & Fleurat-Lessard, F. (2011). Experimental study of acoustic

645

equipment for real-time insect detection in grain bins—Assessment of their potential for

646

infestation risk prediction during long term storage periods. IOBC/wprs Bull, 69, 79-88.

647

46.

Washington, DC: U.S. Patent and Trademark Office.

648 649

Litzkow, C. A., Shuman, D., Kruss, S., & Coffelt, J. A. (1997). U.S. Patent No. 5,646,404.

47.

Magan, N., & Evans, P. (2000). Volatiles as an indicator of fungal activity and

650

differentiation between species, and the potential use of electronic nose technology for

651

early detection of grain spoilage. Journal of Stored Products Research, 36(4), 319-340.

652

48.

Maghirang, E. B., Dowell, F. E., Baker, J. E., & Throne, J. E. (2003). Automated detection

653

of single wheat kernels containing live or dead insects using near–infrared reflectance

654

spectroscopy. Transactions of the ASAE, 46(4), 1277.

655

49.

Manickavasagan, A., Jayas, D. S., & White, N. D. G. (2008). Thermal imaging to detect

656

infestation by Cryptolestes ferrugineus inside wheat kernels. Journal of Stored Products

657

Research, 44(2), 186-192. 28

ACCEPTED MANUSCRIPT

658

50.

Mankin, R. W., Weaver, D. K., Grieshop, M., Larson, B., & Morrill, W. L. (2004).

659

Acoustic system for insect detection in plant stems: comparisons of Cephus cinctus in

660

wheat and Metamasius callizona in bromeliads. J. Agric. Urban Entomol, 21(4), 239-248.

661

51.

Mankin, R. W., Samson, P. R., & Chandler, K. J. (2009). Acoustic detection of

662

Melolonthine larvae in Australian sugarcane. Journal of Economic Entomology, 102(4),

663

1523-1535.

664

52.

Protection, 263.

665 666

Mankin, R., & Hagstrum, D. W. (2012). Acoustic Monitoring of Insects. Stored Product

53.

Mankin, R. W., & Moore, A. (2010). Acoustic detection of Oryctes rhinoceros

667

(Coleoptera: Scarabaeidae: Dynastinae) and Nasutitermes luzonicus (Isoptera: Termitidae)

668

in palm trees in urban Guam. Journal of Economic Entomology, 103(4), 1135-1143.

669

54.

http://clas.mq.edu.au/speech/acoustics/frequency/analog_digital.html

670 671

Mannell, R. (2008). Speech acoustics.

55.

Minkevich, J. M., Demianyk, C. J., White, N. D. G., Jayas, D. S., & Timlick, B. (2002). A

672

rapid method to detect Cryptolestes ferrugineus (Coleoptera: Cucujidae) larvae in stored

673

grain. Canadian Journal of Plant science, 82(3), 591-597.

674

56.

Miranda, J. L., Gerardo, B. D., & Tanguilig III, B. T. (2014). Pest detection and extraction

675

using

676

Communication Engineering, 3(3), 189.

677

57.

processing

techniques. International

Journal

of

Computer

and

Mohan, S. (2007). Ecofriendly post-harvest technologies for management of stored grain insects. Green Farming 1, 45-47.

678 679

image

58.

Mohan, S. (2014). Eco-friendly post-harvest technologies for management of stored grain

680

insects at farm, home and rehouse - a step towards clean grain (insect free) revolution.

681

Department of Cotton, Centre of Plant Breeding and Genetics, TNAU, Coimbatore

682

59.

Mohan, S., Gopalan, M., Sundarababu, P. C., & Sreenarayanan, V. V. (1994). Practical

683

studies on the use of light traps and bait traps in the management of Rhyzopertha dominica

684

(F.) in rice warehouses. International Journal of Pest Management, 40(2), 148-152.

685 686

60.

Mohan, S., & Rajesh, A. (2016). Tools for stored product insects management and technology transfer. Indian Journal of Entomology, 78(special), 59-63.

29

ACCEPTED MANUSCRIPT

687

61.

food grains quality during storage. Internat. J. Current Agric. Res, 1(7), 34-41.

688 689

Nanje Gowda, N. A., & Alagusundaram, K. (2013). Use of thermal imaging to improve the

62.

Narendra, V. G. & Hareesh, K. S. (2010). Prospects of Computer Vision Automated

690

Grading and Sorting Systems in Agricultural and Food Products for Quality Evaluation - A

691

Review. International Journal of Computer Applications, 2(1): 43-65.

692

63.

techniques for stored-product insects in grain. Food Control, 18(2), 157-162.

693 694

Neethirajan, S., Karunakaran, C., Jayas, D. S., & White, N. D. G. (2007). Detection

64.

Niu, Y., Hardy, G., Agarwal, M., Hua, L., & Ren, Y. (2016). Characterization of volatiles

695

Tribolium castaneum (H.) in flour using solid phase micro extraction–gas chromatography

696

mass spectrometry (SPME–GCMS). Food Science and Human Wellness, 5(1), 24-29.

697

65.

Njoroge, A. W., Affognon, H., Mutungi, C., Rohde, B., Richter, U., Hensel, O., & Mankin,

698

R. W. (2016). Frequency and time pattern differences in acoustic signals produced by

699

Prostephanus truncatus (Horn)(Coleoptera: Bostrichidae) and Sitophilus zeamais

700

(Motschulsky)(Coleoptera: Curculionidae) in stored maize. Journal of Stored Products

701

Research, 69, 31-40.

702

66.

Njoroge, A. W., Affognon, H., Mutungi, C., Richter, U., Hensel, O., Rohde, B., & Mankin,

703

R. W. (2017). Bioacoustics of Acanthoscelides obtectus (Coleoptera: Chrysomelidae:

704

Bruchinae) on Phaseolus vulgaris (Fabaceae). Florida Entomologist, 100(1), 109-115.

705

67.

31-43.

706 707

Oerke, E. C. (2006). Crop losses to pests. The Journal of Agricultural Science, 144(01),

68.

Parfitt, J., Barthel, M., & Macnaughton, S. (2010). Food waste within food supply chains:

708

quantification and potential for change to 2050. Philosophical Transactions of the Royal

709

Society B: Biological Sciences, 365(1554), 3065-3081.

710

69.

Patel, K. K., Kar, A., Jha, S. N., & Khan, M. A. (2012). Machine vision system: a tool for

711

quality inspection of food and agricultural products. Journal of Food Science and

712

Technology, 49(2), 123-141.

713

70.

Pearson, T. C., Brabec, D. L., & Schwartz, C. R. (2003). Automated detection of internal

714

insect infestations in whole wheat kernels using a PERTEN SKCS 4100. Applied

715

Engineering in Agriculture, 19(6), 727– 733.

30

ACCEPTED MANUSCRIPT

716

71.

Pearson, T., & Brabec, D. L. (2007). Detection of wheat kernels with hidden insect

717

infestations with an electrically conductive roller mill. Applied Engineering in

718

Agriculture, 23(5), 639.

719

72.

Pederson, J. R. (1992). Insects: identification, damage and detection. In Storage of Cereal

720

Grains and Their Products (pp. 435–489). St. Paul, MN: American Association of Cereal

721

Chemists Inc.

722

73.

Pereira, P. R. V. S., Lazzari, F. A., Lazzari, S. M. N., & Almeida, A. A. (1994, April).

723

Comparison between two methods of insect sampling in stored wheat. In Proceedings of

724

the 6th International Working Conference on Stored-Product Protection (pp. 435-438).

725

74.

Perez-Mendoza, J., Throne, J. E., Dowell, F. E., & Baker, J. E. (2003). Detection of insect

726

fragments in wheat flour by near-infrared spectroscopy. Journal of Stored Products

727

Research, 39(3), 305-312.

728

75.

selection for machine olfaction. Sens. Actuator B Chem. 145:507–515.

729 730

Phaisangittisagul E., Nagle H.T. & Areekul V. (2010). Intelligent method for sensor subset

76.

Pittendrigh, B. R., Huesing, J. E., Shade, R. E., & Murdock, L. L. (1997). Monitoring of

731

rice weevil, Sitophilus oryzae, feeding behavior in maize seeds and the occurence of

732

supernumerary molts in low humidity conditions. Entomologia Experimentalis et

733

Applicata, 83(2), 225-231.

734

77.

Potamitis, I., Ganchev, T., & Kontodimas, D. (2009). On automatic bioacoustic detection

735

of pests: the cases of Rhynchophorus ferrugineus and Sitophilus oryzae. Journal of

736

Economic Entomology, 102(4), 1681-1690.

737

78.

CRC Press.

738 739

79. 80.

744

Ridgway, C., & J. Chambers. 1996. Detection of external and internal insect infestation in wheat by near-infrared reflectance spectroscopy. J. Sci. Food Agric. 71(2):251-264.

742 743

Rajendran, S. (2005). Detection of insect infestation in stored foods. Advances in Food and Nutrition Research, 49, 163-232.

740 741

Prabhu, K. M. M. (2013). Window functions and their applications in signal processing.

81.

Sarwar, M. (2013). Development and boosting of integrated insect pests management in stored grains. Research and Reviews: J Agric Allied Sci, 2(4), 16-20.

31

ACCEPTED MANUSCRIPT

745

82.

Sashidhar, R. B., Ramakrishna, Y., & Bhat, R. V. (1992). Moulds and mycotoxins in

746

sorghum stored in traditional containers in India. Journal of Stored Products

747

Research, 28(4), 257-260.

748

83.

Infested with common storage insects. Cereal Chemistry 73, 744-750.

749 750

Seitz, L.M. & Saucer, D.B., 1996. Volatile compounds and odors in grain sorghum

84.

Semple, R. L. (1980). Inspection procedures for grain handling facilities and methods for

751

detecting stored grain insects. Inspection and detection methods for storage insect pests.

752

FAO Corporate Document Repository.

753

85.

Senthilkumar, T., Jayas, D. S., White, N. D. G., Freund, M. S., Shafai, C., & Thomson, D.

754

J. (2012). Characterization of volatile organic compounds released by granivorous insects

755

in stored wheat. Journal of Stored Products Research, 48, 91-96.

756

86.

India. Journal of Post-Harvest Technology, 2(1), 12-24.

757 758

87.

Shimoda, M., & Honda, K. I. (2013). Insect reactions to light and its applications to pest management. Applied Entomology and Zoology, 48(4), 413-421.

759 760

Sharon, M., Abirami, C. V., & Alagusundaram, K. (2014). Grain storage management in

88.

Shuman, D., & EPSKY, N. D. (2001). Commercialization of the Electronic Grain Probe

761

Insect Counter. In Proceeding of an International Conference on Controlled Atmosphere

762

and Fumigation in Stored Poducts, Fresno, CA (Vol. 29, pp. 665-677).

763

89.

science 42(4): 186–189.

764 765

Singh M, Kaur G, & Kaur S (2004) Machine vision: an upcoming realm Everyman’s

90.

Stetter, J.R., Findlay Jr, M.W., Schroeder, K.M., Yue, C., Penrose, W.R., 1993. Quality

766

classification of grain using a sensor array and pattern recognition. Analytica ChimicaActa

767

284, 1-11.

768

91.

Academic Press.

769 770

92. 93.

775

Wilson A.D. (2012). Review of electronic-nose technologies and algorithms to detect hazardous chemicals in the environment. Proc. Technol. 1:453–463.

773 774

Vithu, P., and Moses, J. A. (2016). Machine vision system for food grain quality evaluation: A review. Trends in Food Science & Technology, 56, 13-20.

771 772

Sun, D. W. (2016). Computer vision technology for food quality evaluation (Ed.).

94.

Wilson, A. D. (2013). Diverse applications of electronic-nose technologies in agriculture and forestry. Sensors, 13(2), 2295-2348. 32

ACCEPTED MANUSCRIPT

776

95.

World Bank (2011). World Bank, FAO, NRI Missing Food: the Case of Post-harvest Grain

777

Losses in Sub-Saharan Africa. Economic Sector Work Report No. 60371-AFRWorld

778

Bank, Washington, DC.

779

96.

Wu, J., Jayas, D. S., Zhang, Q., White, N. D. G., & York, R. K. (2013). Feasibility of the

780

application of electronic nose technology to detect insect infestation in wheat. Canadian

781

Biosystems Engineering, 55.

782

97.

Xing, J., & Guyer, D. (2008). Comparison of transmittance and reflectance to detect insect

783

infestation in Montmorency tart cherry. Computers and Electronics in Agriculture, 64(2),

784

194-201.

785

98.

Yacob Y, Ahmad H, Saad P, Aliana R, Raof A, & Ismail S (2005) A comparison between

786

X-ray and MRI in the postharvest non-destructive detection method. Proceedings of the

787

International Conference on Information Technology and Multimedia at UNITEN (ICIMU

788

‘05), Malaysia.

789 790

99.

Zayas, I. Y., & Flinn, P. W. (1998). Detection of insects in bulk wheat samples with machine vision. Transactions of the ASAE, 41, 883–888.

791

100. Zhang, H., Wang, J., Tian, X., Yu, H., & Yu, Y. (2007). Optimization of sensor array and

792

detection of stored duration of wheat by electronic nose. Journal of Food

793

Engineering, 82(4), 403-408.

33

5% 20% 45%

30%

Weeds

1 2

Insects

Disease

Others

Figure 1: Globally total annual loss in agricultural produce (Oerke, 2006)

3 4

Lid

5

Inner Perforated Cylinder Outer cylinder

Perforated Cone

Perforated Bottom Insect collection bowl

Insect Collection

6 7

Lid

(a)

(b)

8 9 10

1

11 12 Hanger

13

Hopper Outer Cylinder

14

UV Light

15 Perforated Cylinder

Reflector

16 17

Outlet for Clean Grain

Outlet for insects

Motor

18 19

Cone

(c)

Insect collection Bowl

20

(d)

21 22

Figure 2: (a) Indicator device, (b) TNAU automatic insect removal bin, (c) Insect Egg Removal Device, (d) Ultra Violet Light (d)

23

Trap (with permission from Mohan et al., 1994)

24

2

Feature extraction or interpretation Image store

System controlling and decision making

Digitizer

25 26

Figure 3 : Block diagram of a machine vision system (adopted from Singh et al., 2004 and Patel et al., 2012)

27

3

28

Figure 4: X-ray images of pecans with different visible traits. (a) Good nut; (b) insect damage to one cotyledon from inside; (c)

29

visible insect hole; (d) insect damage and insect (with permission from Kotwaliwale et al., 2007a)

30

31 32

Figure 5: Grayscale representation of pseudo color images of non-fungal samples at (a) 0 s, (b) 180 s, (c) 210 s and fungal samples at

33

(d) 0 s, (e) 180 s, (f) 210 s.

34 35 36 37 4

38 39 40 41 Amplifier

42

Digitizer

43

Filter

44 Impulse Converter

45 46 47 48

Adjust detectable threshold Sensors Computer

Documentation

49

Figure 6: Schematic representation of the acoustical probe and signal processing system for detection of insect infestation in grains

50

(adopted from Fleurat-Lessard, 2006)

51

5

Table-1: Characters for insect inspection in storage structures

1

Character

Specification

C - Clear or none

Number of insects

No insects

Require protection from cross-infestation and regular inspection.

F - Few or light

Irregular occurrence of few numbers of < 20 insects per 90 kg sieved sample for a few notations insects. Absence of insects in sacks

(requires disinfestation in near future). 20-300 insects per 90 kg sieve sample for light notation.

MN - Moderate numbers

Regular occurrence and formation of small 50-300 insects per 90 kg sieved sample. population of insects

LN - Large numbers

Large number of insects creeping on the 300-1500 insects per 80 kg sieved sample. stack surface

VLN

-

numbers

Very

large Intense occurrence of insects, audible and >1500 insects per 90 kg sieved sample. dead skins seen around the stalk (Source: Semple, 1980)

1

Table-2: Advantages and limitations of modern methods of insect detection in stored food grains Insect Detection Advantages

Limitations

References

Methods 1 Conductance based method 1.1 Electrically

Suitable for detection of hidden Time-consuming as it inspects a single grain, Brabec et al. 2012,

conductive roller mill

infestation, inexpensive

not

suitable for large capacity, unable to 2017;

Pearson

and

detect the egg, larvae stages and dead Brabec, 2007 insects, not suitable for low moisture content sample 2 Olfactory based methods Dynamic headspace increases Costly, detects only adults insects and not Laopongsit,

et al.,

2.1 Solid phase micro the sensitivity, high sensitivity

suitable for immature insects, requires a 2014; Niu et al., 2016

extraction skilled person Automatic,

non-destructive, Expensive, necessity of long-term training of Magan

and

Evans,

2.2 Electronic Nose suitable for hidden infestation device, use of complicated data fusion 2000; Wu et al.,2013; 2

and mold, rapid, in-situ, suitable techniques, expensive sensors affected by Zhou and Wang, 2011 for early detection

environmental factors and need replacement after some time, cannot detect all species of insects

3.

Electromagnetic spectrum based methods

3.1 Imaging methods Suitable for identification and Expensive, inadequate to detect the dead and Zayas and Flinn, 1998; classification of varieties, insect internal insects, unable to classify the species Miranda et al., 2014, 3.1.1

Machine

Vision infestation, grain discoloration, of insects

Vithu

used for grading of agricultural

2016

and Moses,

within visible range produce Direct method, non-destructive, Inadequate to detect insect egg, high cost, Karunakaran high 3.1.2 X-ray imaging

accuracy,

adequate

to requires

skilled

detect the internal and external measures to operate insects, capable to detect both live and dead insects 3

workers,

need

et

al.,

safety 2003; Kotwaliwale et al., 2007

Suitable to detect all the stages Time-consuming, unable to categorize the Nanje

Gowda

and

of insects, identify the infested mixed variety of grains, the camera is Alagusundaram, 2013 3.1.3 Thermal Imaging and uninfested grains

expensive, cannot identify the development stages of insects

3.2 Non-imaging methods Automatic, 3.2.1

Electronic

real-time Sophisticated system requires a skilled Shuman and Epsky,

Grain monitoring, suitable for bulk person to operate, expensive, unable to 2001; Flinn

et

al.,

Probe Insect Counter storage at any depth

detect the dead insects

2009

Rapid method, detect hidden High cost and required trained person, Neethirajan 3.2.2 NIR Spectroscopy

insect infestation

inadequate to detect low levels of infestation, 2007; requires calibration and care of equipment

4

et

Maghirang

al., et

al., 2003

Acoustic based method Non-destructive,

4.1 Acoustic Detection

detect

internal

automatic, Cannot detect eggs and dead insects, requires Eliopoulos et al., 2015; and

external sound

and

vibration-insulated

structure, Fleurat-Lessard et al.

insects, high sensitivity, suitable detects within a suitable range, sophisticated 2006; Mankin, 2012;

4

for taking reliable decision in equipment silos, estimate density of insects 2

5

Pearson et al. 2007