Development of Text-Reading Camera to Assist the Legally Blind Using Neural Network Image Processing Algorithm

Eyesight defects are one of the most difficult human sense failures that happen to a person. It caused struggles in reading, walking, and performing other activities that require vision. This study developed a text-reading camera with the application of a neural network image processing algorithm that will assist the legally blind in reading product labels and codes, receipts, currencies, product package labels, and other objects with text related labels. The proponents employ quantitative research design in gathering data, measurable metrics for evaluation of the developed prototype, and software development life cycle method`s principles and characteristics to achieve the goal. The proponents created an actual device that reads texts within the image and transforms them into voice with the utilization of Convolution Neural Network Image Processing Algorithm for image enhancement and quality corrections. The developed device undergoes numerous tests to ensure that the objectives of the study are really attained. The developed device was tested, validated, and evaluated by the respondents whose vision profile consists of partially sighted, normal vision, and legal blindness. Results show that device functional suitability, reliability, and performance efficiency were rated by the respondents with a grand mean equivalent to verbal interpretation of “agree,” while usability and portability criterion got the lowest grand mean equivalent to verbal interpretation of “slightly agree” which means that the developed tool is useful and still commendable yet can be concluded that more innovation in the hardware components used and software application developed must be adhered to.