- Email: [email protected]
Objective measurement of veal color for classification purposes
Meat Science 53 (1999) 203±209
www.elsevier.com/locate/meatsci
Objective measurement of veal color for classi®cation purposes C. Denoyelle*, F. Berny Institut de l'Elevage, Service Viande, 149 rue de Bercy, 75595 Paris cedex 12, France Received 4 November 1998; received in revised form 15 April 1999; accepted 15 April 1999
Abstract 2300 veal calves were used to compare the ability of chromameters to measure the veal meat color on-line and to develop a relationship between instrumental and visual assessments to predict the color score according to the EC-system of classi®cation. The meat color was assessed subjectively by 3 trained judges and objectively by 2 chromameters CR300 and CR310, 45 min post mortem on the Rectus abdominis, on the external side, skin having been removed. R2 values reached 0.70 for the CR300 and 0.75 for the CR310. The equations of prediction classi®ed correctly up to 87% of carcasses. These data indicate that chromameters (Minolta CR310 and CR300) can be used on-line to measure objectively veal meat color at the end of the slaughterline. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction The color of veal is an important quality trait: for the consumers, it is the most important ®rst impression when they buy the meat. Moreover in France, veal meat color is important to breeders because it contributes to the carcass price. At present, evaluation of veal color for classi®cation purposes is done subjectively, at the end of slaughterline, using dierent people, dierent muscles and dierent illumination conditions. This evaluation is based on the EC-system with 4 color classes (1=white, 2=light pinkish, 3=pinkish, 4=red). This classi®cation is often the cause of commercial con¯icts between breeders and the slaughterhouse. Hence, there is an urgent need in French veal meat processing to supplement the EC-system of classi®cation with an objective system of veal color measurement. Several methods exist to measure meat color in the laboratory (Eikelenboom, Bolink, & Hulsegge, 1990; Renerre, 1981; Renerre, 1988; Swatland, 1995): including determination of haem pigment content (Hornsey, 1956). Although these methods are accurate, they are not useful to assess veal color on-line at the end of the slaughterline. * Corresponding author Tel.: +33-1-40-04-53-02; fax: +33-1-40-0449-60. E-mail address: [email protected] (C. Denoyelle)
Eikelenboom (1989) concluded that the Minolta chromameter II could be used to develop a veal color classi®cation system based on objective measurements. Becherel (1991) compared the ability of 5 instruments (the Minolta chromameter CR300, the Sensoptic colorimeter, the re¯ectometer Retrolux III and optical ®bres from Sensoptic and the Danish Meat Research Institute) to measure the color on 367 veal carcasses. The results showed that the Minolta chromameter CR300 was the best instrument to predict veal color classi®cation. Since 1991, new instruments have been developed, in particular, the Minolta CR310 which diers from the CR300 in the area of surface measured (CR300: diameter 8 mm, CR310: 50 mm). A previous study (Denoyelle & Jabet 1997), compared the ability of these new instruments and the CR300 to measure objectively beef meat color. Repeatability, reproducibility and ease of use in industrial conditions were estimated. The results con®rmed the ability of CR300 and CR310 to measure meat color. They gave the best results (estimated with R2 values) and were the most adapted for use in industrial conditions. In the light of these results, the chromameter CR310 could be expected to measure veal meat color successfully. Thus, the objective of this experiment was to compare the ability of Minolta chromameters CR300 and CR310 to measure veal meat color on-line and to develop a relationship between instrumental and visual assessment using to the EC-system of classi®cation.
0309-1740/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(99)00056-X
204
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
2. Materials and methods 2.1. Animals 2300 veal calves from dierent breeds, age, and sex were used. They were killed in 6 slaughterhouses (S1±S6) from the west of France. Fat (5 classes, from 1=lean to 5=fat) and conformation (E.U.R.O.P classi®cation: 5 classes, from E = ``good'' to P = ``bad'' conformation) scores according to the EC-system and trimmed carcass weights were recorded at about 45 min post mortem. 2.2. Visual assessment Four training sessions were organized to select judges on their repeatability and reproducibility. During each session, 2 series of measurements were performed. The sample order was changed at random between the 2 series. By series, the people judged the meat color of 60 Rectus abdominis. In total, 60 muscles 2 series 8 people 4 sessions i.e. 3840 color scores were done. From the 8 trained people, 3 were selected. Visual assessments were performed at the end of the slaughterline (around 45 min post mortem) independently by the 3 selected judges on the Rectus abdominis and the whole carcass according to the EC-system of classi®cation of veal color (OFIVAL, 1976). 4 classes were used: 1=white, 2=light pinkish, 3=pinkish, 4=red. At the same time, the color score given by the slaughterhouse judge was recorded. 2.3. Physical measurements Color measurements were carried out with Minolta chromameters CR300 and CR310 at 45 min post mortem on the Rectus abdominis, on the external side, skin having been removed. Results were expressed as L (lightness), a (redness) and b (yellowness) in the CIELAB system, with the D65 lightsource (Cassens et al., 1995). The a value is a measure of a color continuum from red to green, and b value is a measure of a color continuum from yellow to blue. Greater L value denotes lighter meat and greater a and b values indicate a more red and yellow color, respectively. For each veal carcass, 3 measurements were performed on dierent places on the surface of the muscle. 2.4. Statistical analysis Multiple stepwise regression analysis was carried out by SAS (1988), between the EC-system of classi®cation (score from 1 to 4) and L, a, b values. For each carcass, the average of the 3 measurements (visual and physical assessment) was used. Equations of prediction were developed with 80% of carcasses and tested with the remaining 20%, from 2
samples. The ®rst one was the sample (G1) with all veal carcasses (2300 carcasses), the second sample (G2) was a part of G1 where the 3 judges had done the same visual color score for each carcass (1650 carcasses). Performance of the chromameters was estimated by the R2 value and the percentage of carcasses correctly graded. This percentage represented carcasses whose score predicted by the chromameter corresponded to the score given by visual assessment. 3. Results and discussion Carcass characteristics are given in Table 1. On average, the carcasses weighed 120.65 kg, had a conformation score ``O'' and a fatness score ``3''. Color score distributions in the 6 slaughterhouses are shown in Table 2. Of the 2300 veal carcasses measured, only 4.4% gave a score of ``4'' and 8% a score of ``1''. The majority (87.1%) gave scores of ``2'' and ``3''. This distribution re¯ects the diculty breeders have in producing veal meat with a white color (score 1) and their eorts to avoid red colored veal (score 4). In France, the carcass is Table 1 Carcasses characteristics (n=2300)
Carcass weight (kg) Conformation scorec Fat scoreb
Mean
SDa
Min
Max
120.65 0 (9.3) 3 (7.7)
16.35 2.12 0.90
54.8 P (5) 1 (2)
183.3 E(17) 4(11)
a
SD: standard deviation. Fat score: 5 classes from 1=lean to 5=fat. c Conformation score: 5 classes from E=``good'' con®rmation to P=``bad'' conformation. b
Table 2 Color scores distribution (n=2300 carcasses) Color score
a
Sl
S2 S3 S4 S5 S6 Total a
n= % n= % n= % n= % n= % n= % n= %
S: slaughterhouse.
1
2
3
4
32 11.59 32 5.27 11 4.40 1 0.24 66 11.85 48 18.82 190 8.0
175 63.41 327 53.87 141 56.40 173 40.80 369 66.25 129 50.59 1314 55.5
62 22.46 228 37.56 90 36.00 212 50.00 106 19.03 63 24.71 761 32.1
7 2.54 20 3.29 8 3.20 38 8.96 16 2.87 15 5.88 104 4.4
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
less valuable with a score of ``4'' as consumers prefer white or light pinkish meat. The relationship between average lightness L, redness a and yellowness b and the color scores in each slaughterhouse (S1, S2, S3, S4, S5, S6) are shown in Figs.1±6 for the CR310 and the CR300. Lightness decreased from the white meat (score ``1'') to the red meat (score ``4''). Redness logically increased with the color score and yellowness decreased. The relationships between L, a, b values and color scores were the same for the 2 chromameters. However, the results between slaughterhouses were more consistent with the CR310, except for the b value where the results were very different. Dierences between slaughterhouses could be partly explained by the inequal distribution of color scores (see Table 2) and the more consistent results of
205
the CR310 by the larger diameter of the head (50 mm vs 8 mm for the CR300). Correlations between L, a, b values and visual assessment are presented in Table 3. There was a relatively strong correlation between L and a values (r=ÿ0.64 and r=ÿ0.69) respectively for CR300 and CR310, but only between L and b values (r=0.44) for the CR310; for the CR300, this correlation was not signi®cant. Regarding the correlations between L, a, b values and visual assessment of the 3 trained judges, they were relatively strong, around 0.6, except for the b value from the CR300. Table 4 shows the results of multiple stepwise regressions. Ten dierent models (M1±M10) were studied (5 per chromameter). For similar models (with the same variables), correlations were higher for CR310 than for
Fig 1. Relationship between average L value and color score for each slaughterhouse (S1±S6) - CR300.
Fig. 2. Relationship between average L* value and color score for each slaughterhouse (S1±S6)±CR310.
206
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
Fig. 3. Relationship between average a value and color score for each slaughterhouse (S1±S6) - CR300.
Fig. 4. Relationship between average a value and color score for each slaughterhouse (S1±S6) - CR310.
CR300: e. g. 8 points between M3 and M8, 5 points between M5 and M10. R2 values reached 0.7 with CR300 and 0.75 with CR310. The introduction in the model of quadratic terms for the variables L, a, b did not improve R2 values to any major extent, 0.02 point for the CR300 between M3 and M5, 0.01 point for the
CR310 between M8 and M10. In fact, the relationship between Lab values and color score seemed to be linear. Nevertheless, in order to provide the best accuracy, models M5 (CR300) and M10 (CR310) which included L2, a2 and b2 were selected to develop and test equations of prediction.
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
207
Fig. 5. Relationship between average b value and color score for each slaughterhouse (S1±S6) - CR300.
Fig. 6. Relationship between average b value and color score for each slaughterhouse (S1±S6) - CR310.
Table 3 Correlations (r) between L*, a*, b* and visual assessment from trained judges CR300
a*CR300 b*CR300 a*CR310 b*CR310 J1a J2a J3a MJb a b c
CR310
L*
a*
b*
L*
a*
b*
ÿ0.64 0.02 n.s. ± ± ÿ0.64 ÿ0.59 ÿ0.66 ÿ0.65
± ÿ0.02 n.s.c ± ± 0.65 0.61 0.65 0.65
± ± ± ± ÿ0.33 ÿ0.34 ÿ0.34 ÿ0.34
± ± ÿ0.69 0.44 ÿ0.78 ÿ0.72 ÿ0.78 ÿ0.78
± ± ± ÿ0.21 0.58 0.51 0.56 0.57
± ± ± ± ÿ0.53 ÿ0.53 ÿ0.54 ÿ0.54
J1, J2, J3: the 3 trained judges. MJ: mean color score given by the 3 trained judges. n.s.: no signi®cance
208
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
Table 4 Multiple stepwise regression between instrumental measurements and visual assessment (trained judges and slaughterhouse judges) Variables
R2 (T)a
R2 (S)b
Variables
R2 (T)
R2 (S)
a* (M1) a*, b* (M2) a*, b*, L* (M3) a*, b*, L*, L2* (M4) a*, b*, L*, L2*, b2* (M5)
0.47 0.59 0.68 0.69 0.70
0.25 0.31 0.34 0.36 0.36
L* (M6) L*, b* (M7) L*, b*, a* (M8) L*, b*, a*, L2* (M9) L*, b*, a*, L2*, b2 (M10)
0.67 0.73 0.74 0.74 0.75
0.32 0.35 0.36 0.38 0.39
a b
T: trained judges. S: slaughterhouse judges.
Table 5 Percentage of carcasses correctly classi®ed with chromameters CR300 and CR310 from models (M5) and (M10)-sample G1 (2300 carcasses)a
Table 6 Percentage of carcasses correctly classi®ed with chromameters CR300 and CR310 from models (M5) and (M10)-samples G2 (1650 carcasses)a
Draws
Draws
1 2 3 4 5 6 7 8 9 10 Mean
Carcasses correctly graded (%) CR300
CR310
78.0 74.8 77.9 74.7 74.8 74.0 75.3 78.2 77.2 74.6 75.9
83.5 79.0 80.5 78.5 82.2 81.0 81.2 84.9 79.5 78.5 80.8
a Equations of prediction were developed with 80% of carcasses and tested with the remaining 20%.
The correlations between physical measurements and slaughterhouse visual assessment were low (around 0.4). These lower results could be explained because only a single judge was used and it was not the same person in each slaughterhouse (they are though trained and controlled by an external body), In addition the conditions of slaughterhouse assessment were ``commercial'' and not technical. Tables 5 and 6 show the percentage of carcasses correctly classi®ed with the chromameters CR300 and CR310. With G1, 75.9% of carcasses were correctly classi®ed with the CR300 and 80.8% with the CR310. With G2, 82.4% of carcasses were correctly classi®ed with the CR300 and 87.6% with the CR310. By color score, the percentage of carcasses correctly classi®ed was lower (around 20±40 points) with color scores ``1'' and ``4''. These results were very acceptable and higher than those obtained by a single judge. In fact, a trained judge classi®ed correctly 75±79% of carcasses against another judge. These tests con®rmed the superiority of the CR310 to predict veal color, but this superiority was relatively low, so the use of CR300 is still possible. These results suggest that chromameters can predict the veal color score according to the EC-system of classi®cation.
1 2 3 4 5 6 7 8 9 10 Mean
Carcasses correctly graded (%) CR300
CR310
81.5 80.2 85.2 82.5 83.2 79.1 81.4 84.7 82.1 84.3 82.4
89.6 86.4 85.5 86.9 89.0 88.0 83.8 91.2 86.9 88.9 87.6
a Equations of prediction were developed with 80% of carcasses and tested with the remaining 20%.
Finally, 2 equations of prediction were proposed: For the CR300 : color score 18:850 ÿ 0:675 L 0:173 a ÿ 0:188 b 0:006 L 2 0:008 b 2 For the CR 310 : color score 23:560 ÿ 0:753 L ÿ 0:277 b 0:006 L 2 0:003 a2 0:02 b2 For clarity not all decimals are quoted but they are necessary to use these equations with accuracy. Completed equations can be sent on request. Practically, in slaughterhouses, these results could be used to measure objectively the color of veal meat. Two uses are possible. The ®rst would compute the color score by means of equations of prediction on a computer at the end of slaughterline. This solution would give a result immediately and could be printed on a ticket, with the weight, the conformation score and the fat score. The second would take the more representative variables (L and a for CR300, L and b for CR310) to build an abacus. The result is less accurate but could be used easily on the slaughterline.
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
4. Conclusion Chromameters CR310 and CR300 were useful to predict veal meat color on-line. There were strong correlations between L, a, b values and visual assessment by trained judges. However, further investigations currently in progress will probably improve the accuracy of prediction, especially for the extreme color classes (1=white and 4=red), and will help consolidate the equations of prediction. Acknowledgements This work was supported by OFIVAL and Interveaux. References BeÂcherel, F. (1991). Projet europeÂen de test de dieÂrents appareils pour mesurer la couleur des carcasses de veau sur la chaiÃne d'abattage. Ed. C.I.V, Paris.
209
Cassens R.G., Demeyer D., Eikelenboom G., Honikel K.O., Johansson G., Nielsen T., Renerre M., Richardson I. & Sakata R. (1995). Recommendation of reference method for assessment of meat color. Proceedings of the 41st International Congress of Meat Science and Technology, San Antonio. Denoyelle, C., Jabet, S. (1997). Objective measurement of beef meat color. Proceedings of the 43rd International Congress of Meat Science and Technology, Auckland. Eikelenboom, G. (1989). The assessment of veal colour for classi®cation purposes. Proceedings of the 35th International Congress of Meat Science and Technology, Copenhagen. Eikelenboom, G., Bolink, A. H., & Hulsegge, B. (1990). Evaluation d'appareils invasifs pour la mesure de la couleur de la viande de veau en vue de la classi®cation. Viandes et Produits CarneÂs, 11(6), 246±247. Hornsey, H. C. (1956). The color of cooked cured pork. Journal of the Science of Food and Agriculture, 7, 534±540. OFIVAL (1976). Classi®cation EUROP des Veaux. Ed. OFIVAL, Paris Renerre, M. (1976). La couleur de la viande et sa mesure. Viandes et Produits CarneÂs, 2(5), 10±16. Renerre M. Quelles recommandations pour mesurer la couleur de la viande au laboratoire? Industries Agro Alimentaire, juin 530-533 SAS, (1988). SAS/STAT User's guide, release 6.03. Ed. SAS Institute, Cary, NC, USA Swatland, H.J. (1995). In on-line evaluation of meat. Ed. Technomic, Basel 185±206
www.elsevier.com/locate/meatsci
Objective measurement of veal color for classi®cation purposes C. Denoyelle*, F. Berny Institut de l'Elevage, Service Viande, 149 rue de Bercy, 75595 Paris cedex 12, France Received 4 November 1998; received in revised form 15 April 1999; accepted 15 April 1999
Abstract 2300 veal calves were used to compare the ability of chromameters to measure the veal meat color on-line and to develop a relationship between instrumental and visual assessments to predict the color score according to the EC-system of classi®cation. The meat color was assessed subjectively by 3 trained judges and objectively by 2 chromameters CR300 and CR310, 45 min post mortem on the Rectus abdominis, on the external side, skin having been removed. R2 values reached 0.70 for the CR300 and 0.75 for the CR310. The equations of prediction classi®ed correctly up to 87% of carcasses. These data indicate that chromameters (Minolta CR310 and CR300) can be used on-line to measure objectively veal meat color at the end of the slaughterline. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction The color of veal is an important quality trait: for the consumers, it is the most important ®rst impression when they buy the meat. Moreover in France, veal meat color is important to breeders because it contributes to the carcass price. At present, evaluation of veal color for classi®cation purposes is done subjectively, at the end of slaughterline, using dierent people, dierent muscles and dierent illumination conditions. This evaluation is based on the EC-system with 4 color classes (1=white, 2=light pinkish, 3=pinkish, 4=red). This classi®cation is often the cause of commercial con¯icts between breeders and the slaughterhouse. Hence, there is an urgent need in French veal meat processing to supplement the EC-system of classi®cation with an objective system of veal color measurement. Several methods exist to measure meat color in the laboratory (Eikelenboom, Bolink, & Hulsegge, 1990; Renerre, 1981; Renerre, 1988; Swatland, 1995): including determination of haem pigment content (Hornsey, 1956). Although these methods are accurate, they are not useful to assess veal color on-line at the end of the slaughterline. * Corresponding author Tel.: +33-1-40-04-53-02; fax: +33-1-40-0449-60. E-mail address: [email protected] (C. Denoyelle)
Eikelenboom (1989) concluded that the Minolta chromameter II could be used to develop a veal color classi®cation system based on objective measurements. Becherel (1991) compared the ability of 5 instruments (the Minolta chromameter CR300, the Sensoptic colorimeter, the re¯ectometer Retrolux III and optical ®bres from Sensoptic and the Danish Meat Research Institute) to measure the color on 367 veal carcasses. The results showed that the Minolta chromameter CR300 was the best instrument to predict veal color classi®cation. Since 1991, new instruments have been developed, in particular, the Minolta CR310 which diers from the CR300 in the area of surface measured (CR300: diameter 8 mm, CR310: 50 mm). A previous study (Denoyelle & Jabet 1997), compared the ability of these new instruments and the CR300 to measure objectively beef meat color. Repeatability, reproducibility and ease of use in industrial conditions were estimated. The results con®rmed the ability of CR300 and CR310 to measure meat color. They gave the best results (estimated with R2 values) and were the most adapted for use in industrial conditions. In the light of these results, the chromameter CR310 could be expected to measure veal meat color successfully. Thus, the objective of this experiment was to compare the ability of Minolta chromameters CR300 and CR310 to measure veal meat color on-line and to develop a relationship between instrumental and visual assessment using to the EC-system of classi®cation.
0309-1740/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(99)00056-X
204
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
2. Materials and methods 2.1. Animals 2300 veal calves from dierent breeds, age, and sex were used. They were killed in 6 slaughterhouses (S1±S6) from the west of France. Fat (5 classes, from 1=lean to 5=fat) and conformation (E.U.R.O.P classi®cation: 5 classes, from E = ``good'' to P = ``bad'' conformation) scores according to the EC-system and trimmed carcass weights were recorded at about 45 min post mortem. 2.2. Visual assessment Four training sessions were organized to select judges on their repeatability and reproducibility. During each session, 2 series of measurements were performed. The sample order was changed at random between the 2 series. By series, the people judged the meat color of 60 Rectus abdominis. In total, 60 muscles 2 series 8 people 4 sessions i.e. 3840 color scores were done. From the 8 trained people, 3 were selected. Visual assessments were performed at the end of the slaughterline (around 45 min post mortem) independently by the 3 selected judges on the Rectus abdominis and the whole carcass according to the EC-system of classi®cation of veal color (OFIVAL, 1976). 4 classes were used: 1=white, 2=light pinkish, 3=pinkish, 4=red. At the same time, the color score given by the slaughterhouse judge was recorded. 2.3. Physical measurements Color measurements were carried out with Minolta chromameters CR300 and CR310 at 45 min post mortem on the Rectus abdominis, on the external side, skin having been removed. Results were expressed as L (lightness), a (redness) and b (yellowness) in the CIELAB system, with the D65 lightsource (Cassens et al., 1995). The a value is a measure of a color continuum from red to green, and b value is a measure of a color continuum from yellow to blue. Greater L value denotes lighter meat and greater a and b values indicate a more red and yellow color, respectively. For each veal carcass, 3 measurements were performed on dierent places on the surface of the muscle. 2.4. Statistical analysis Multiple stepwise regression analysis was carried out by SAS (1988), between the EC-system of classi®cation (score from 1 to 4) and L, a, b values. For each carcass, the average of the 3 measurements (visual and physical assessment) was used. Equations of prediction were developed with 80% of carcasses and tested with the remaining 20%, from 2
samples. The ®rst one was the sample (G1) with all veal carcasses (2300 carcasses), the second sample (G2) was a part of G1 where the 3 judges had done the same visual color score for each carcass (1650 carcasses). Performance of the chromameters was estimated by the R2 value and the percentage of carcasses correctly graded. This percentage represented carcasses whose score predicted by the chromameter corresponded to the score given by visual assessment. 3. Results and discussion Carcass characteristics are given in Table 1. On average, the carcasses weighed 120.65 kg, had a conformation score ``O'' and a fatness score ``3''. Color score distributions in the 6 slaughterhouses are shown in Table 2. Of the 2300 veal carcasses measured, only 4.4% gave a score of ``4'' and 8% a score of ``1''. The majority (87.1%) gave scores of ``2'' and ``3''. This distribution re¯ects the diculty breeders have in producing veal meat with a white color (score 1) and their eorts to avoid red colored veal (score 4). In France, the carcass is Table 1 Carcasses characteristics (n=2300)
Carcass weight (kg) Conformation scorec Fat scoreb
Mean
SDa
Min
Max
120.65 0 (9.3) 3 (7.7)
16.35 2.12 0.90
54.8 P (5) 1 (2)
183.3 E(17) 4(11)
a
SD: standard deviation. Fat score: 5 classes from 1=lean to 5=fat. c Conformation score: 5 classes from E=``good'' con®rmation to P=``bad'' conformation. b
Table 2 Color scores distribution (n=2300 carcasses) Color score
a
Sl
S2 S3 S4 S5 S6 Total a
n= % n= % n= % n= % n= % n= % n= %
S: slaughterhouse.
1
2
3
4
32 11.59 32 5.27 11 4.40 1 0.24 66 11.85 48 18.82 190 8.0
175 63.41 327 53.87 141 56.40 173 40.80 369 66.25 129 50.59 1314 55.5
62 22.46 228 37.56 90 36.00 212 50.00 106 19.03 63 24.71 761 32.1
7 2.54 20 3.29 8 3.20 38 8.96 16 2.87 15 5.88 104 4.4
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
less valuable with a score of ``4'' as consumers prefer white or light pinkish meat. The relationship between average lightness L, redness a and yellowness b and the color scores in each slaughterhouse (S1, S2, S3, S4, S5, S6) are shown in Figs.1±6 for the CR310 and the CR300. Lightness decreased from the white meat (score ``1'') to the red meat (score ``4''). Redness logically increased with the color score and yellowness decreased. The relationships between L, a, b values and color scores were the same for the 2 chromameters. However, the results between slaughterhouses were more consistent with the CR310, except for the b value where the results were very different. Dierences between slaughterhouses could be partly explained by the inequal distribution of color scores (see Table 2) and the more consistent results of
205
the CR310 by the larger diameter of the head (50 mm vs 8 mm for the CR300). Correlations between L, a, b values and visual assessment are presented in Table 3. There was a relatively strong correlation between L and a values (r=ÿ0.64 and r=ÿ0.69) respectively for CR300 and CR310, but only between L and b values (r=0.44) for the CR310; for the CR300, this correlation was not signi®cant. Regarding the correlations between L, a, b values and visual assessment of the 3 trained judges, they were relatively strong, around 0.6, except for the b value from the CR300. Table 4 shows the results of multiple stepwise regressions. Ten dierent models (M1±M10) were studied (5 per chromameter). For similar models (with the same variables), correlations were higher for CR310 than for
Fig 1. Relationship between average L value and color score for each slaughterhouse (S1±S6) - CR300.
Fig. 2. Relationship between average L* value and color score for each slaughterhouse (S1±S6)±CR310.
206
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
Fig. 3. Relationship between average a value and color score for each slaughterhouse (S1±S6) - CR300.
Fig. 4. Relationship between average a value and color score for each slaughterhouse (S1±S6) - CR310.
CR300: e. g. 8 points between M3 and M8, 5 points between M5 and M10. R2 values reached 0.7 with CR300 and 0.75 with CR310. The introduction in the model of quadratic terms for the variables L, a, b did not improve R2 values to any major extent, 0.02 point for the CR300 between M3 and M5, 0.01 point for the
CR310 between M8 and M10. In fact, the relationship between Lab values and color score seemed to be linear. Nevertheless, in order to provide the best accuracy, models M5 (CR300) and M10 (CR310) which included L2, a2 and b2 were selected to develop and test equations of prediction.
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
207
Fig. 5. Relationship between average b value and color score for each slaughterhouse (S1±S6) - CR300.
Fig. 6. Relationship between average b value and color score for each slaughterhouse (S1±S6) - CR310.
Table 3 Correlations (r) between L*, a*, b* and visual assessment from trained judges CR300
a*CR300 b*CR300 a*CR310 b*CR310 J1a J2a J3a MJb a b c
CR310
L*
a*
b*
L*
a*
b*
ÿ0.64 0.02 n.s. ± ± ÿ0.64 ÿ0.59 ÿ0.66 ÿ0.65
± ÿ0.02 n.s.c ± ± 0.65 0.61 0.65 0.65
± ± ± ± ÿ0.33 ÿ0.34 ÿ0.34 ÿ0.34
± ± ÿ0.69 0.44 ÿ0.78 ÿ0.72 ÿ0.78 ÿ0.78
± ± ± ÿ0.21 0.58 0.51 0.56 0.57
± ± ± ± ÿ0.53 ÿ0.53 ÿ0.54 ÿ0.54
J1, J2, J3: the 3 trained judges. MJ: mean color score given by the 3 trained judges. n.s.: no signi®cance
208
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
Table 4 Multiple stepwise regression between instrumental measurements and visual assessment (trained judges and slaughterhouse judges) Variables
R2 (T)a
R2 (S)b
Variables
R2 (T)
R2 (S)
a* (M1) a*, b* (M2) a*, b*, L* (M3) a*, b*, L*, L2* (M4) a*, b*, L*, L2*, b2* (M5)
0.47 0.59 0.68 0.69 0.70
0.25 0.31 0.34 0.36 0.36
L* (M6) L*, b* (M7) L*, b*, a* (M8) L*, b*, a*, L2* (M9) L*, b*, a*, L2*, b2 (M10)
0.67 0.73 0.74 0.74 0.75
0.32 0.35 0.36 0.38 0.39
a b
T: trained judges. S: slaughterhouse judges.
Table 5 Percentage of carcasses correctly classi®ed with chromameters CR300 and CR310 from models (M5) and (M10)-sample G1 (2300 carcasses)a
Table 6 Percentage of carcasses correctly classi®ed with chromameters CR300 and CR310 from models (M5) and (M10)-samples G2 (1650 carcasses)a
Draws
Draws
1 2 3 4 5 6 7 8 9 10 Mean
Carcasses correctly graded (%) CR300
CR310
78.0 74.8 77.9 74.7 74.8 74.0 75.3 78.2 77.2 74.6 75.9
83.5 79.0 80.5 78.5 82.2 81.0 81.2 84.9 79.5 78.5 80.8
a Equations of prediction were developed with 80% of carcasses and tested with the remaining 20%.
The correlations between physical measurements and slaughterhouse visual assessment were low (around 0.4). These lower results could be explained because only a single judge was used and it was not the same person in each slaughterhouse (they are though trained and controlled by an external body), In addition the conditions of slaughterhouse assessment were ``commercial'' and not technical. Tables 5 and 6 show the percentage of carcasses correctly classi®ed with the chromameters CR300 and CR310. With G1, 75.9% of carcasses were correctly classi®ed with the CR300 and 80.8% with the CR310. With G2, 82.4% of carcasses were correctly classi®ed with the CR300 and 87.6% with the CR310. By color score, the percentage of carcasses correctly classi®ed was lower (around 20±40 points) with color scores ``1'' and ``4''. These results were very acceptable and higher than those obtained by a single judge. In fact, a trained judge classi®ed correctly 75±79% of carcasses against another judge. These tests con®rmed the superiority of the CR310 to predict veal color, but this superiority was relatively low, so the use of CR300 is still possible. These results suggest that chromameters can predict the veal color score according to the EC-system of classi®cation.
1 2 3 4 5 6 7 8 9 10 Mean
Carcasses correctly graded (%) CR300
CR310
81.5 80.2 85.2 82.5 83.2 79.1 81.4 84.7 82.1 84.3 82.4
89.6 86.4 85.5 86.9 89.0 88.0 83.8 91.2 86.9 88.9 87.6
a Equations of prediction were developed with 80% of carcasses and tested with the remaining 20%.
Finally, 2 equations of prediction were proposed: For the CR300 : color score 18:850 ÿ 0:675 L 0:173 a ÿ 0:188 b 0:006 L 2 0:008 b 2 For the CR 310 : color score 23:560 ÿ 0:753 L ÿ 0:277 b 0:006 L 2 0:003 a2 0:02 b2 For clarity not all decimals are quoted but they are necessary to use these equations with accuracy. Completed equations can be sent on request. Practically, in slaughterhouses, these results could be used to measure objectively the color of veal meat. Two uses are possible. The ®rst would compute the color score by means of equations of prediction on a computer at the end of slaughterline. This solution would give a result immediately and could be printed on a ticket, with the weight, the conformation score and the fat score. The second would take the more representative variables (L and a for CR300, L and b for CR310) to build an abacus. The result is less accurate but could be used easily on the slaughterline.
C. Denoyelle, F. Berny / Meat Science 53 (1999) 203±209
4. Conclusion Chromameters CR310 and CR300 were useful to predict veal meat color on-line. There were strong correlations between L, a, b values and visual assessment by trained judges. However, further investigations currently in progress will probably improve the accuracy of prediction, especially for the extreme color classes (1=white and 4=red), and will help consolidate the equations of prediction. Acknowledgements This work was supported by OFIVAL and Interveaux. References BeÂcherel, F. (1991). Projet europeÂen de test de dieÂrents appareils pour mesurer la couleur des carcasses de veau sur la chaiÃne d'abattage. Ed. C.I.V, Paris.
209
Cassens R.G., Demeyer D., Eikelenboom G., Honikel K.O., Johansson G., Nielsen T., Renerre M., Richardson I. & Sakata R. (1995). Recommendation of reference method for assessment of meat color. Proceedings of the 41st International Congress of Meat Science and Technology, San Antonio. Denoyelle, C., Jabet, S. (1997). Objective measurement of beef meat color. Proceedings of the 43rd International Congress of Meat Science and Technology, Auckland. Eikelenboom, G. (1989). The assessment of veal colour for classi®cation purposes. Proceedings of the 35th International Congress of Meat Science and Technology, Copenhagen. Eikelenboom, G., Bolink, A. H., & Hulsegge, B. (1990). Evaluation d'appareils invasifs pour la mesure de la couleur de la viande de veau en vue de la classi®cation. Viandes et Produits CarneÂs, 11(6), 246±247. Hornsey, H. C. (1956). The color of cooked cured pork. Journal of the Science of Food and Agriculture, 7, 534±540. OFIVAL (1976). Classi®cation EUROP des Veaux. Ed. OFIVAL, Paris Renerre, M. (1976). La couleur de la viande et sa mesure. Viandes et Produits CarneÂs, 2(5), 10±16. Renerre M. Quelles recommandations pour mesurer la couleur de la viande au laboratoire? Industries Agro Alimentaire, juin 530-533 SAS, (1988). SAS/STAT User's guide, release 6.03. Ed. SAS Institute, Cary, NC, USA Swatland, H.J. (1995). In on-line evaluation of meat. Ed. Technomic, Basel 185±206