Image Enhancement, Restoration and Segmentation Methods for Vision Based Sign Language Recognition

Abstract

SLR (Sign Language Recognition) plays an essential part in increasing the communication for the deaf and speechless people. Because Sign Language is the sole language used by the deaf and speechless to exchange messages with one another. Sign Language is entirely distinct from generally spoken languages because It possesses its own grammar. But still SLR, or sign language recognition, is the most difficult area for practitioners and researchers. The SLR is mostly required because of its ability to remove the communication barrier for deaf and dumb people. To overcome this problem this research presents the comprehensive review of different studies based on sign language which includes various techniques of Artificial Intelligence (AI) used for vision based SLR for human understanding that have employed in recent past which will help to develop a model to fill the communication gap. The main objective of this work is comparison of image enhancement techniques (Histogram Equalization, Contrast Limited Adaptive Histogram Equalization), image restoration techniques (Mean Filter, Gaussian Filter) and image segmentation techniques (Edge Based, Region Based and Cluster Based) by applying on images and also discuss their scope of area and limitations. Important SLR challenges like database limitations and regional sign languages have also been covered. This paper aim is to help researcher to know about presented methods of sign language translations along with new findings, so that a new system can be developed with more accuracy, easy to use and comfortable for deaf people. Remaining part of this paper will contain the literature review and different approaches of sign language recognition and different techniques used for vision-based approach of sign language recognition.