- Email: [email protected]
A reading system for the blind based on geometrical shapes
Compur. E&c. Vol. 12. No. 2. pp. 311-320. Pnnted in Great Britain
0360-I3I5 88 53.00+ 0.00
1988
Pergamon
MAHMOUD
Press
plc
HABIB*
Department of Electrical and Computer Engineering. College of Engineering and Petroleum, Kuwait University, P.O. Box 5969, 13060 Safat. Kuwait (Received 27 November 1985; amended 14 December 1986) Abstract-This paper illustrates an alternative optically based system for the purpose of reading textual information for the visually impaired. The textual alphabet character information, in particular, Arabic and/or English is first converted to geometrical shape symbols. These symbols are optically simple to recognize. Scanners are used to produce the corresponding signals for the processor system for determination of the alphabetic character and produce the appropriate signals to be used as inputs to the voice synthesis system. A special keyboard as an input peripheral for entering the geometrical shape set into a computer is presented. The proposed system facilitates to the blind student a textual leading machine in a multilanguage environment as well as being an alternative to Braille code representation.
INTRODUCTION
The visually handicapped use the braille notation as a common media for accessing written texual information instead of the black print used by the sighted person. The braille notation is an arrangement of bubbles, 2 x 3 or 2 x 4 matrix, raised in heavy paper or plastic sheets. Each arrangement of these bubbles corresponds to an alphabet, numbers and other textual information. The production of braille coded books, periodical and newspapers, is accomplished manually or automatically by special printing offices located at libraries or government establishments. The conversion of text to braille is expensive and limited in terms of the amount of information a blind person acquires compared with a sighted person. Reading braille is accomplished by movement of fingers over the raised bubbles and recognizing the arrangement and its corresponding textual information, i.e. the alphabets. Thus, reading braille depends entirely on hand motion and finger sensitivity. Consequently, a motion handicapped or dyslexic blind cannot benefit from braille. Furthermore, an Arabic blind student who is learning English or English blind student learning Arabic, needs to acquire the capability to read braille representation of Arabic and English languages. An automatic braille reading system has been reported which mechanically reads and converts braille to ASCII code. However, this system does not produce synthesized speech. Thus, utilization of computers by the blind requires that input and output peripherals, printers and readers, must be capabie of handiing braiiie format. The advancement in very large scale integration (VLSI) chip manufacturing and computer technologies has progressed rapidly in terms of compatibility and processing speed. The availability of computer based speech synthesis systems has become economically feasible for persons to acquire for home applications. Speech synthesis systems produce concatenated basic speech units phoneme. Text-to-speech systems apply letter to sound rules known as synthesis by rule[2]. Through programming techniques, concatenation of phonemes, some speech synthesis systems have been adopted for multilanguage capability[3]. Image processing techniques have improved considerably and systems have been produced to aid the visually impaired as well as the dyslexic student to acquire texual information. As an example, the Kurzweil English text reading machine for the blind is now available in some libraries[4]. However, this machine is expensive, difficult to access and adds financial burden for personal use at home. A microcomputer reading aid for the blind student has been designed by Omatayo which relies on text already stored in the memory of the computer[5].
*Presentaddress: Visiting Professor, University of Maryland, College Park. MD 20742, U.S.A. 311
MAHMO~~D
31’
HABIB
An Arab blind student definitely needs a machine capable of reading texual information both in Arabic and English and producing synthetic speech in both languages. Although several attempts have been reported in Arabic character recognition, there is no reading machine in existence for rl_ *__!-I_ I___.____r~, 1Ilc ArdtJlG lilIl&l~~C L3J In this paper, a geometrical shape coded character set (GSCCS) is introduced and employed as an alternative to braille code representation of textual information. It is shown that GSCCS can be utilized in an Arabic/English, bilingual environment, The system consists of a processor and input/output peripherals. The input peripherals are: (a) optical scanners and detection hardware; (b) a geometrical shape coded keyboard; (c) ASCII code keyboard (Arabic/English). The output peripharals are: (a) bilingual speech synthesizer; (b) geometrical shape code printer; (c) Arabic/English character printer. SYSTEM
DESIGN
The system design objectives for building a reading system for the blind student can be summarized as: (1) establish a character set for the representation of the alphabet ehich can be processed and recognized without employing sophisticated image processing techniques; (2) capability of the reading system in a bilingual application environment; (3) the reading system should be organized around a computer, preferably a microprocessor with software flexibility to change from one language to the other without changing the input devices; (4) economically competitive. The reading system for the blind student in a bilingual environment is based on a bilingual geometrical shape representation of the alphabets. The printed geometrical shapes are scanned and recognized. The output of the scanners are fed into a processor to produce a command, based on the letter recognized, to a speech synthesizer to produce the appropriate sound. The geometrical shapes are produced by programming the processor to convert an input from an ASCII keyboard to printed geometrical shapes corresponding to the letter pressed on the keyboard. A specially geometrically shapes based keyboard is designed and shown later in this paper. This keyboard is utilized as an input device to the processor. The block diagram of the reading system is shown in Fig. 1. The output peripherals are: speech synthesizer system, text printer and geometrical shape printer. Upon the desired application by the user; the system operates on the mode desired, i.e. read geometrical shapes and produce speech. Geometrical shape coded character set (GSCCS)
The geometrical shape coded character set (GSCCS) consists of a set of universal symbols selected to represent the alphabets, namely circle, square, and triangle. These shapes are relatively
Binary
Optical scanners
Rinted CSC
-
and
signal -
detection
The p-ocessor
c
Text printer
w
GSC pnnter
hardware A
GSC based
A
Keyboard
keyboard
Fig. 1. Schematic diagram for the reading system.
A reading
system
for the blind
313
simpe to recognize optically compared with the alphabet characters[6]. The geometrical shapes contain variations to encompass a class of existing alphabet and the phonetic symbols of a language. The variation of the basic geometrical symbols are accomplished inside the boundary f _, or rne shape by shades and iines. Attributes are added outside the geometricai shapes; dots and dashes for extra flexibility. The development of the GSCCS is accomplished by first starting with the basic geometrical shape; second, a variation inside the shape is selected, i.e. shades (fully or partially) and lines (single or double); third, the kind of attribute is selected for outside the geometrical shape, i.e. dots or horizontal dashes stacked vertically. This is shown in Fig. 2. Each geometrical shape and its variation forms a subset of coded shapes which can represent a class of 30 distinct elements in a particular group of similar characteristics. Thus, this superset can represent three major subsets for a total of 90 elements. Extensions of the scheme to encompass more elements can be easily accomplished by the addition to another geometrical shape as an attribute. The sets of the geometrical shape are shown in Fig. 2. Table 1 depicts the set of the geometrical shapes and the corresponding phonemes. The symbols on the left of the shape correspond to the Arabic phonemes and to the right the English phonemes. Starting with the Arabic, which has 6 vowel phonemes, the correspondng English vowels are selected and then matched with set of circles of the GSCCS. Since the circle set of GSCCS can accommodate 30 phonemes and since English has 12 vowels, the remaining 6 vowels are listed in the second column of the circle set. Further, the diphthongs remain in the same circle set. The consonant phonemes of both languages are matched and accordingly assigned to the squares in the GSCCS. There are some consonants in Arabic which do not exist in English and vice versa[7]. These particular phonemes are moved to the triangular shapes of the GSCCS. Thus, the relationship between text information, i.e. the alphabet and/or phonemes and the geometrical shape coping is established. This relationship is utilized to facilitate an alternative method to convert text, Arabic or English to speech using a simple recognition scheme. The text alphabet of the language is first converted to the appropriate geometrical code then printed. The geometrical shapes are passed through a scanner which converts them to binary code. The scanning method and the electronic circuitry has been illustrated[6]. Generation of GSCC Since printed GSCC are needed as input to the scanner system, the geometrical shapes are produced by two methods mainly: (a) from an ASCII keyboard; (b) from a specially designed GSC keyboard. The program flow chart shown in the following section illustrates the production of printed GSCC from an ASCII keyboard. Upon pressing of a key, the ASCII binary representation of the letter is detected by the processor (see Fig. l), and an output geometrical shape is printed equivalent to the letter. The output is 9 x 9 pixel print for each GSCC. The maximum print is fifty GSC characters or attributes per line. The program was written in Basic and used on an HP-86 microprocessor with a dot matrix printer. This keyboard and this program enable a typist without . -^-training to type text directiy to the processor and produce printed tiscc which can be utiiized by the blind to read using the optical scanners, detection hardware and converted to speech. Note that, this keyboard employs one language, either English or Arabic.
0
Empty
l
Fully
Partially
@
0
., 0
0
-,
0
shaded shaded
aI
Single
aD
DoubLe line
:,0
i
=, 0 g
lme
Dot attributes Line
attribute
Fig. 2. The sets of geometrical
shapes.
311
MAHMOUD HABIB
Table I. Geometrical shaoes for Arabic and Ennlish ahonemes
me m: qi m-
m= t-gA.; AZM mi AA= A; GSC keyboard A multilanguage keyboard is developed for the geometrical shape schemes which can be used as an input device to a computer. The design layout of the keyboard with the geometrical shapes as keys is shown in Fig. 3. It consists of two parts. The keys or buttons are the geometrical shapes which correspond to the phoneme or alphabet of the language. The dials shown (top) are the adaptive mechanism for multilanguage use of the keys. One dial is used for language selection.
Lonquoge seletion
6
2 4
I
Space keys
Fig. 3. GSC based keyboard.
I
A reading
system
for the blind
315
Optional dials can be included for functions such as pitch control. accent and typing style. The GSC keyboard can be connected as an input device to a computer where the output of the keyboard is appropriately coded in binary. Since the ASCII code representation of the English and Arabic languages has been standardized, this code is developed for the GSC representation in binary numbers. Table 2 shows the ASCII representation of the GSC. The binary numbers transmitted to the computer are processed using the program flow chart shown in the following section to generate printed GSCC. Needless to say that the keyboard also produces commands for speech synthesis. This keyboard is a prototype which does not include all typing characters; however, it can be utilized for phonetic input directly to the computer, To use this keyboard, the geometrical shape key is pressed first, then the attribute key. The advantage of this keyboard is that it can be utilized for several languages. SYSTEM
OPERATION
The basic operation of the reading system for the blind in a bilingual environment, Arabic and English, is the capability to recognize a simple geometrical shape coded character set as a representation of the alphabet and/or phoneme. There are several input-output modes in which the reading system operate. A microprocessor, Intel 8085, based system’s block diagram is shown in Fig. 4. The switches S, SZS3 S, (top) select the mode of operation, i.e. reading GSCC, keyboard.
Table 2. ASCII rep~~ntatio~
of geometrical shapes code
Controi umt
Fig. 4.
Microprocessor
system block diagram.
Figure 5a depicts the Flow Chart of the operation of the system and the service routine for each mode. The tirst mode is reading GSCC. The printed GSCC are presented to the scanners. The scanner system hardware re~o~ni2es each shape and an appropriate binary code, s-bit word, represen~tion of the equivalent shape is produ~d. The &bit word signals are the input to the processor which performs the tasks of receiving and decoding. The software program’s Aow chart for the subroutine to load the &bit code into the accumulator is shown in Fig. 5b. A command procedure is performed to the output device, i.e. speech synthesizer system, based on the alphabet recognized by the processor to produce the appropriate sound. Based on the required output such as printing a GSCC or an alphabet, the processor will produce the appropriate command to the output devices desired. The second mode of operation is receiving input from an ASCII keyboard and/or from the GSCC based keyboard. Through software (Fig. SC), the processor produces the appropriate output desired. For example, printed GSCC from ASCII keyboard or printed GSCC from GSCC based keyboard, It was intended that the reading system by flexible and not restricted to a particular processor, Le. Intel, National semiconductors etc. Accordingly, the program was written in Pascal on an HP 64~ development system. The various input~utput modes were examined and varified. Although the speech synthesizer was not integrated into the system due to the complexity of applying ietter-to-sound rules, the GSCC Hardware recognizer and the GSCC keyboard were constructed and used as part of the system. The incorporation of the letter-to-sound rules in our reading system is left for future research. The assembly program for the Intel 8085 with the Optical scanners, Detection Hardware, keyboard and GSC keyboard was obtained. The system produces: (a) command to speech synthesis system (simulated); (b) printed GSC; (c) printed text,
A reading system for the blind
Yes Read code
switch from
accumulator
1 Disable interrupt fll&l flops
Fig. S(a) (cupion on p. 319)
317
318
Produce error message
Send address
f Perform speech synthesis routine Yes 1 Perform printing routine
Pwlice
text
emble
sqnai
1 -
Yes
No
II
Fig. .5(a)-continued
Perform GSC printer routine
-c
Load dota code
I
:o accuwwitor
Fig. 5(b)
319
A reading system for the blind
Fed I/P Port for cobmn check Yes
1
I
Increment L decrement column cwnter
Set row counter
r
1
r A0 A
MweLto accumulator I
0
Fig. 5(c)
Fig. 5. (a) System’s operation flow chart. (b) GSC binary signal service routine. (c) Keyboard service routine.
CONCLUSION This paper proposed an optically based system for reading Arabic and English textual information for the blind student. The Arabic alphabetical character set is mapped into a set of geometrical shapes, namely circle, square and triangle. This mapping of the alphabetical character set to the geometrical shapes facilitates an alternative solution to the problem of optical character recognition. The system consisting of scanners, keyboard and a geometrical based keyboard to a speech synthesizer and processor produces an output speech or alphabet or geometrical shape printer. The proposed system provides an intermediate solution to the problem of optical recognition of the oddly shaped characters of the Arabic language without using sophisticated image processing techniques. This is because the recognition process is applied to simple to recognize geometrical shapes. This facilitates an economical reading aid for the blind student in a bilingual environment, namely Arabic and English. The system described provide an alternative
310
MAHMOLJD HABIB
to braille and completely aids the blind to read (with synthesized voice) text converted to geometrical shapes. Furthermore, the system enables the blind to communicate with other blind persons, i.e. by sending a letter, by printed geometrical shape representation of texual information. Another advantage is that a sighted person can transmit texual information to a blind person through a normal type setting keyboard by utilization of text to geometrical shape representation and conversion. REFERENCES I. Braggeman F. and Van Spronsen, A braille reading system. Euromicro Conference (September 1985). 2. Flanagan J. er al., Synthetic voices for computers. IEEE-Spectrum, p. 22 (October 1970). 3. Mahjoub A., Arabic system software requirements. Computer Processing of the Arabic Language Wtbrkshop, Vol. 2, Kuwait (April 1985). 4. Kurzweil R. C., Kurzweil reading machine for the blind. Proceedings of Johns Hopkins First Sational Search for Application of Personal Computing to Aid the Handicapped, pp. 220-223. (October 1981). 5. Omotayo 0.. A microcomputer-based reading aid for blind students. IEEE Trans. Educ. E-26, No. 1 (Yovemixr 1983). 6. Habib M. er al., New geometrical shape code and its optical recognition scheme. Int. J. Elecfron. 1, No. 1 (July 1986). 7. Habib M., The characteristics of arabic speech vowels. J. Univ. Kuwait (Science) (July 1986).
0360-I3I5 88 53.00+ 0.00
1988
Pergamon
MAHMOUD
Press
plc
HABIB*
Department of Electrical and Computer Engineering. College of Engineering and Petroleum, Kuwait University, P.O. Box 5969, 13060 Safat. Kuwait (Received 27 November 1985; amended 14 December 1986) Abstract-This paper illustrates an alternative optically based system for the purpose of reading textual information for the visually impaired. The textual alphabet character information, in particular, Arabic and/or English is first converted to geometrical shape symbols. These symbols are optically simple to recognize. Scanners are used to produce the corresponding signals for the processor system for determination of the alphabetic character and produce the appropriate signals to be used as inputs to the voice synthesis system. A special keyboard as an input peripheral for entering the geometrical shape set into a computer is presented. The proposed system facilitates to the blind student a textual leading machine in a multilanguage environment as well as being an alternative to Braille code representation.
INTRODUCTION
The visually handicapped use the braille notation as a common media for accessing written texual information instead of the black print used by the sighted person. The braille notation is an arrangement of bubbles, 2 x 3 or 2 x 4 matrix, raised in heavy paper or plastic sheets. Each arrangement of these bubbles corresponds to an alphabet, numbers and other textual information. The production of braille coded books, periodical and newspapers, is accomplished manually or automatically by special printing offices located at libraries or government establishments. The conversion of text to braille is expensive and limited in terms of the amount of information a blind person acquires compared with a sighted person. Reading braille is accomplished by movement of fingers over the raised bubbles and recognizing the arrangement and its corresponding textual information, i.e. the alphabets. Thus, reading braille depends entirely on hand motion and finger sensitivity. Consequently, a motion handicapped or dyslexic blind cannot benefit from braille. Furthermore, an Arabic blind student who is learning English or English blind student learning Arabic, needs to acquire the capability to read braille representation of Arabic and English languages. An automatic braille reading system has been reported which mechanically reads and converts braille to ASCII code. However, this system does not produce synthesized speech. Thus, utilization of computers by the blind requires that input and output peripherals, printers and readers, must be capabie of handiing braiiie format. The advancement in very large scale integration (VLSI) chip manufacturing and computer technologies has progressed rapidly in terms of compatibility and processing speed. The availability of computer based speech synthesis systems has become economically feasible for persons to acquire for home applications. Speech synthesis systems produce concatenated basic speech units phoneme. Text-to-speech systems apply letter to sound rules known as synthesis by rule[2]. Through programming techniques, concatenation of phonemes, some speech synthesis systems have been adopted for multilanguage capability[3]. Image processing techniques have improved considerably and systems have been produced to aid the visually impaired as well as the dyslexic student to acquire texual information. As an example, the Kurzweil English text reading machine for the blind is now available in some libraries[4]. However, this machine is expensive, difficult to access and adds financial burden for personal use at home. A microcomputer reading aid for the blind student has been designed by Omatayo which relies on text already stored in the memory of the computer[5].
*Presentaddress: Visiting Professor, University of Maryland, College Park. MD 20742, U.S.A. 311
MAHMO~~D
31’
HABIB
An Arab blind student definitely needs a machine capable of reading texual information both in Arabic and English and producing synthetic speech in both languages. Although several attempts have been reported in Arabic character recognition, there is no reading machine in existence for rl_ *__!-I_ I___.____r~, 1Ilc ArdtJlG lilIl&l~~C L3J In this paper, a geometrical shape coded character set (GSCCS) is introduced and employed as an alternative to braille code representation of textual information. It is shown that GSCCS can be utilized in an Arabic/English, bilingual environment, The system consists of a processor and input/output peripherals. The input peripherals are: (a) optical scanners and detection hardware; (b) a geometrical shape coded keyboard; (c) ASCII code keyboard (Arabic/English). The output peripharals are: (a) bilingual speech synthesizer; (b) geometrical shape code printer; (c) Arabic/English character printer. SYSTEM
DESIGN
The system design objectives for building a reading system for the blind student can be summarized as: (1) establish a character set for the representation of the alphabet ehich can be processed and recognized without employing sophisticated image processing techniques; (2) capability of the reading system in a bilingual application environment; (3) the reading system should be organized around a computer, preferably a microprocessor with software flexibility to change from one language to the other without changing the input devices; (4) economically competitive. The reading system for the blind student in a bilingual environment is based on a bilingual geometrical shape representation of the alphabets. The printed geometrical shapes are scanned and recognized. The output of the scanners are fed into a processor to produce a command, based on the letter recognized, to a speech synthesizer to produce the appropriate sound. The geometrical shapes are produced by programming the processor to convert an input from an ASCII keyboard to printed geometrical shapes corresponding to the letter pressed on the keyboard. A specially geometrically shapes based keyboard is designed and shown later in this paper. This keyboard is utilized as an input device to the processor. The block diagram of the reading system is shown in Fig. 1. The output peripherals are: speech synthesizer system, text printer and geometrical shape printer. Upon the desired application by the user; the system operates on the mode desired, i.e. read geometrical shapes and produce speech. Geometrical shape coded character set (GSCCS)
The geometrical shape coded character set (GSCCS) consists of a set of universal symbols selected to represent the alphabets, namely circle, square, and triangle. These shapes are relatively
Binary
Optical scanners
Rinted CSC
-
and
signal -
detection
The p-ocessor
c
Text printer
w
GSC pnnter
hardware A
GSC based
A
Keyboard
keyboard
Fig. 1. Schematic diagram for the reading system.
A reading
system
for the blind
313
simpe to recognize optically compared with the alphabet characters[6]. The geometrical shapes contain variations to encompass a class of existing alphabet and the phonetic symbols of a language. The variation of the basic geometrical symbols are accomplished inside the boundary f _, or rne shape by shades and iines. Attributes are added outside the geometricai shapes; dots and dashes for extra flexibility. The development of the GSCCS is accomplished by first starting with the basic geometrical shape; second, a variation inside the shape is selected, i.e. shades (fully or partially) and lines (single or double); third, the kind of attribute is selected for outside the geometrical shape, i.e. dots or horizontal dashes stacked vertically. This is shown in Fig. 2. Each geometrical shape and its variation forms a subset of coded shapes which can represent a class of 30 distinct elements in a particular group of similar characteristics. Thus, this superset can represent three major subsets for a total of 90 elements. Extensions of the scheme to encompass more elements can be easily accomplished by the addition to another geometrical shape as an attribute. The sets of the geometrical shape are shown in Fig. 2. Table 1 depicts the set of the geometrical shapes and the corresponding phonemes. The symbols on the left of the shape correspond to the Arabic phonemes and to the right the English phonemes. Starting with the Arabic, which has 6 vowel phonemes, the correspondng English vowels are selected and then matched with set of circles of the GSCCS. Since the circle set of GSCCS can accommodate 30 phonemes and since English has 12 vowels, the remaining 6 vowels are listed in the second column of the circle set. Further, the diphthongs remain in the same circle set. The consonant phonemes of both languages are matched and accordingly assigned to the squares in the GSCCS. There are some consonants in Arabic which do not exist in English and vice versa[7]. These particular phonemes are moved to the triangular shapes of the GSCCS. Thus, the relationship between text information, i.e. the alphabet and/or phonemes and the geometrical shape coping is established. This relationship is utilized to facilitate an alternative method to convert text, Arabic or English to speech using a simple recognition scheme. The text alphabet of the language is first converted to the appropriate geometrical code then printed. The geometrical shapes are passed through a scanner which converts them to binary code. The scanning method and the electronic circuitry has been illustrated[6]. Generation of GSCC Since printed GSCC are needed as input to the scanner system, the geometrical shapes are produced by two methods mainly: (a) from an ASCII keyboard; (b) from a specially designed GSC keyboard. The program flow chart shown in the following section illustrates the production of printed GSCC from an ASCII keyboard. Upon pressing of a key, the ASCII binary representation of the letter is detected by the processor (see Fig. l), and an output geometrical shape is printed equivalent to the letter. The output is 9 x 9 pixel print for each GSCC. The maximum print is fifty GSC characters or attributes per line. The program was written in Basic and used on an HP-86 microprocessor with a dot matrix printer. This keyboard and this program enable a typist without . -^-training to type text directiy to the processor and produce printed tiscc which can be utiiized by the blind to read using the optical scanners, detection hardware and converted to speech. Note that, this keyboard employs one language, either English or Arabic.
0
Empty
l
Fully
Partially
@
0
., 0
0
-,
0
shaded shaded
aI
Single
aD
DoubLe line
:,0
i
=, 0 g
lme
Dot attributes Line
attribute
Fig. 2. The sets of geometrical
shapes.
311
MAHMOUD HABIB
Table I. Geometrical shaoes for Arabic and Ennlish ahonemes
me m: qi m-
m= t-gA.; AZM mi AA= A; GSC keyboard A multilanguage keyboard is developed for the geometrical shape schemes which can be used as an input device to a computer. The design layout of the keyboard with the geometrical shapes as keys is shown in Fig. 3. It consists of two parts. The keys or buttons are the geometrical shapes which correspond to the phoneme or alphabet of the language. The dials shown (top) are the adaptive mechanism for multilanguage use of the keys. One dial is used for language selection.
Lonquoge seletion
6
2 4
I
Space keys
Fig. 3. GSC based keyboard.
I
A reading
system
for the blind
315
Optional dials can be included for functions such as pitch control. accent and typing style. The GSC keyboard can be connected as an input device to a computer where the output of the keyboard is appropriately coded in binary. Since the ASCII code representation of the English and Arabic languages has been standardized, this code is developed for the GSC representation in binary numbers. Table 2 shows the ASCII representation of the GSC. The binary numbers transmitted to the computer are processed using the program flow chart shown in the following section to generate printed GSCC. Needless to say that the keyboard also produces commands for speech synthesis. This keyboard is a prototype which does not include all typing characters; however, it can be utilized for phonetic input directly to the computer, To use this keyboard, the geometrical shape key is pressed first, then the attribute key. The advantage of this keyboard is that it can be utilized for several languages. SYSTEM
OPERATION
The basic operation of the reading system for the blind in a bilingual environment, Arabic and English, is the capability to recognize a simple geometrical shape coded character set as a representation of the alphabet and/or phoneme. There are several input-output modes in which the reading system operate. A microprocessor, Intel 8085, based system’s block diagram is shown in Fig. 4. The switches S, SZS3 S, (top) select the mode of operation, i.e. reading GSCC, keyboard.
Table 2. ASCII rep~~ntatio~
of geometrical shapes code
Controi umt
Fig. 4.
Microprocessor
system block diagram.
Figure 5a depicts the Flow Chart of the operation of the system and the service routine for each mode. The tirst mode is reading GSCC. The printed GSCC are presented to the scanners. The scanner system hardware re~o~ni2es each shape and an appropriate binary code, s-bit word, represen~tion of the equivalent shape is produ~d. The &bit word signals are the input to the processor which performs the tasks of receiving and decoding. The software program’s Aow chart for the subroutine to load the &bit code into the accumulator is shown in Fig. 5b. A command procedure is performed to the output device, i.e. speech synthesizer system, based on the alphabet recognized by the processor to produce the appropriate sound. Based on the required output such as printing a GSCC or an alphabet, the processor will produce the appropriate command to the output devices desired. The second mode of operation is receiving input from an ASCII keyboard and/or from the GSCC based keyboard. Through software (Fig. SC), the processor produces the appropriate output desired. For example, printed GSCC from ASCII keyboard or printed GSCC from GSCC based keyboard, It was intended that the reading system by flexible and not restricted to a particular processor, Le. Intel, National semiconductors etc. Accordingly, the program was written in Pascal on an HP 64~ development system. The various input~utput modes were examined and varified. Although the speech synthesizer was not integrated into the system due to the complexity of applying ietter-to-sound rules, the GSCC Hardware recognizer and the GSCC keyboard were constructed and used as part of the system. The incorporation of the letter-to-sound rules in our reading system is left for future research. The assembly program for the Intel 8085 with the Optical scanners, Detection Hardware, keyboard and GSC keyboard was obtained. The system produces: (a) command to speech synthesis system (simulated); (b) printed GSC; (c) printed text,
A reading system for the blind
Yes Read code
switch from
accumulator
1 Disable interrupt fll&l flops
Fig. S(a) (cupion on p. 319)
317
318
Produce error message
Send address
f Perform speech synthesis routine Yes 1 Perform printing routine
Pwlice
text
emble
sqnai
1 -
Yes
No
II
Fig. .5(a)-continued
Perform GSC printer routine
-c
Load dota code
I
:o accuwwitor
Fig. 5(b)
319
A reading system for the blind
Fed I/P Port for cobmn check Yes
1
I
Increment L decrement column cwnter
Set row counter
r
1
r A0 A
MweLto accumulator I
0
Fig. 5(c)
Fig. 5. (a) System’s operation flow chart. (b) GSC binary signal service routine. (c) Keyboard service routine.
CONCLUSION This paper proposed an optically based system for reading Arabic and English textual information for the blind student. The Arabic alphabetical character set is mapped into a set of geometrical shapes, namely circle, square and triangle. This mapping of the alphabetical character set to the geometrical shapes facilitates an alternative solution to the problem of optical character recognition. The system consisting of scanners, keyboard and a geometrical based keyboard to a speech synthesizer and processor produces an output speech or alphabet or geometrical shape printer. The proposed system provides an intermediate solution to the problem of optical recognition of the oddly shaped characters of the Arabic language without using sophisticated image processing techniques. This is because the recognition process is applied to simple to recognize geometrical shapes. This facilitates an economical reading aid for the blind student in a bilingual environment, namely Arabic and English. The system described provide an alternative
310
MAHMOLJD HABIB
to braille and completely aids the blind to read (with synthesized voice) text converted to geometrical shapes. Furthermore, the system enables the blind to communicate with other blind persons, i.e. by sending a letter, by printed geometrical shape representation of texual information. Another advantage is that a sighted person can transmit texual information to a blind person through a normal type setting keyboard by utilization of text to geometrical shape representation and conversion. REFERENCES I. Braggeman F. and Van Spronsen, A braille reading system. Euromicro Conference (September 1985). 2. Flanagan J. er al., Synthetic voices for computers. IEEE-Spectrum, p. 22 (October 1970). 3. Mahjoub A., Arabic system software requirements. Computer Processing of the Arabic Language Wtbrkshop, Vol. 2, Kuwait (April 1985). 4. Kurzweil R. C., Kurzweil reading machine for the blind. Proceedings of Johns Hopkins First Sational Search for Application of Personal Computing to Aid the Handicapped, pp. 220-223. (October 1981). 5. Omotayo 0.. A microcomputer-based reading aid for blind students. IEEE Trans. Educ. E-26, No. 1 (Yovemixr 1983). 6. Habib M. er al., New geometrical shape code and its optical recognition scheme. Int. J. Elecfron. 1, No. 1 (July 1986). 7. Habib M., The characteristics of arabic speech vowels. J. Univ. Kuwait (Science) (July 1986).