AUTOMATIC BRAILLE RECOGNITION METHODS: A REVIEW AND TAXONOMY OF MODERN APPROACHES
Keywords:
Keywords: Braille, optical recognition, image processing, machine learning, deep learning, hybrid model, assistive technologies.Abstract
Abstract. This paper provides a systematic review and taxonomy of methods for automatic recognition of Braille writing from digital images. Classical image-processing approaches, traditional machine learning, and modern deep-learning methods are compared. The strengths, weaknesses, application conditions, and suitability of each group to the geometric properties of Braille are examined. Based on the analysis, the superiority of a hybrid (geometric and probabilistic) approach for camera-captured images under variable conditions is justified.
References
1.Majidova G.A. Models and Algorithms for Converting Braille Writing into Audio Information: PhD dissertation. – Tashkent: TUIT, 2025.
2.Gonzalez R.C., Woods R.E. Digital Image Processing. – Pearson, 2018.
3.Goodfellow I., Bengio Y., Courville A. Deep Learning. – MIT Press, 2016.
4.Hartley R., Zisserman A. Multiple View Geometry in Computer Vision. – Cambridge University Press, 2004.
5.Bishop C.M. Pattern Recognition and Machine Learning. – Springer, 2006