A Portable Device for the Translation of Braille to Literary Text n Andrew Pasquale n Curtin University of Technology

Introduction n Purpose of the Device n The Braille System n Prototype operation n The Prototype System –Image Capture –Quadrature Linear Motion Detection –Optical Braille Character Recognition n Future Development

Purpose of the Device n Written communications barrier between seeing and blind person’s –Schooling system –Workplace n Diabetic Peripheral Polyneuropathy (diabetes being the major cause of blindness is Australia) n All specifications and requirements were sourced from the Association for the Blind WA

The Braille System n 6 embossed dots gives 64 possible combinations n Many different grades of braille (ie English, US, Greek, Arabic, math, music etc etc) –each with respective context sensitive contraction rules, is similar to shorthand text n e.g, English Grade 2 – p = ‘people’ – : = ‘wh’ OR ‘which’ OR ‘:’ (colon)

Prototype Block Diagram

Prototype Implementation

Scanner Operation

Image Capture

CCD Operation n 128 pixel linear CCD array n 100 Hz frame rate (SI pulse triggered) n Pixel voltages clocked out serially at 50 KHz

Image Capture Operation

Quadrature Linear Motion Detection n To provide the DSP with reference distance and direction information for the correlation of camera frames with physical braille dots n Provides interrupt pulse signals to the DSP to capture exactly 92 frames per inch from the CCD’s continuous outputs n Information on scanner’s direction of movement allows CCD frames resulting from hand jitter and back scanning to be corrected for. n DSP can then form an accurate 2D image of braille page from a 1D camera system

Theory of Operation n Codewheel (slotted disk) and a pair of IR LED’s / photodiodes from a computer mouse n DSP performs logic

DSP Platform n Texas Instruments TMS320VC5402 DSP Evaluation Board –100 MHz Operation with low power usage –Onboard programmable 22 KHz ADC, DAC –Real time DSP output and debugging through parallel port to host PC –C bases IDE with extensive code library

DSP Operation n Interrupt Driven DSP Software Algorithm: –Wait for codewheel interrupt, is scanner moving forward? n Wait for next SI (camera frame) interrupt (is within 10 ms) –Sample incoming CCD frame data –determine braille cell

Dot Recognition in Frame

Vertical Misalignment n Assuming the linear CCD array will always be correctly aligned to braille dots is an impractical assumption

Vertical Misalignment Correction Technique n Simply perform logical OR operation of successive captured camera frames

Cell Compilation n Create ‘slice’ variable representing the presence of dots within a specific camera frame n correlate right ‘slice’ of braille cell with left ‘slice’ of braille cell to create complete braille cell n the physical ambiguity inherent in single sided braille cells can be compensated for by using simple rules (depending on the type of braille medium) –e.g n is = decimal point, if inbetween two numbers n is = k, if not between two numbers

Cell Compilation n Represent each braille cell by binary braille code n Conversion into various braille types can then use binary code as basis to apply contraction rules

Braille Decompression n Decompression algorithms for converting braille mediums into text are available n eg, algorithm developed by Paul Blenkhorn/Iain Murray n basically a state machine with exception handling functionality (for non-printable contraction characters such as capital/ number indicator etc) n uses switchable look up table that includes the context sensitive contraction rules of the respective braille code (English, Arabic, math etc) being read n Conversion into various braille types can then use binary code as basis to apply contraction rules to

Future Improvements n LCD display or text to speech converter output n Ergonomics: eg four wheel tracking system n Area CCD array n Double as “Text Reader”

Conclusion n Project Achievements –implementation and utilisation of quadrature linear motion detection –image capture of physical braille medium –Computer system able of optical character recognition from a dynamic medium –proven basis for a portable, real time braille to literary text translation device

Conclusion n Once developed, the device has potential to act as a written communication’s gateway between seeing and vision impaired person’s.

Acknowledgements n Academic Supervisor: Iain Murray n Tech Support at ECE Curtin University n John Heppel, Mark Fowler, Ivus Siewart n ???? Any Questions ????