Imprinted Braille-character pattern recognition using image processing techniques

Abstract

The study has proposed a novel and innovative system that uses image processing techniques for converting an imprinted Braille pattern into natural language character. The processing takes place through a series of steps implemented in MATLAB. The major step is the removal of noise from the scanned image of an imprinted Braille pattern and other artifacts and then segmenting the scanned image to yield lines. These segmented lines are then further processed to yield Braille characters. The complete system is fully tested and works efficiently for both manually written as well as computer generated Braille for Urdu language and is found to give accurate results. This novel idea can be implemented for any language, thereby has a profound positive impact on the development of education material for visually challenged individuals together with other applications.

A Review on Software Algorithms for Optical Recognition of Embossed Braille Characters

“…In this study, a statistical method based on Beta distribution to estimate the thresholds was put to use. Segmenting the braille image into lines, and also segmenting the lines into character pattern was proposed by Mohd Wajid et al in [36].Histogram representation of the image was being used for segmentation portrayal. The 3 peaks in the vertical histogram correspond to a sole Braille line.…”

“…The dot position for each cell was then detected using the GFV. Histogram representation was used by Mohd Wajid et al to recognise the Braille dots [36] and this representation for the images was given wherein the length of every peak was proportional to the number of dots present in the line. The minimum separation between the peaks was equal to the vertical separation between the dots and further, the maximum width of the peak was equal to the vertical dimension of the dot.…”

E-Braille Documents: Novel Method For Error Free Generation

“…3: For the vectorQ, compute the Number of Null Projections (N N P ) in the space R X and the number of zero elements inQ, will be equal to N N P . 4: Perform rotation transformation by δθ in anticlockwise direction and crop the unwanted boundary region [6] and store the result in the same variable I , mathematically we can say…”

“…The method proposed in [6] Algorithm 1 Algorithm for Estimation ofθ 1: For the binary complemented and cropped image matrix I (Fig. 3, 7 and 9) compute the following…”

“…In the era of internet and e-books there is a dire need for In [6] the author has developed a method for automatic recognition of Braille pattern from its Braille image with many applications; like, teaching/learning Braille to visually challenged children, for machine driven correction of examination script in Braille, for secret communication and also in creation of e-library for Braille. Creation of images, either by scanning or photography, may have rotational distortion, which make the decoding of these patterns difficult.…”