Bangla sign language interpretation using image processing

Abstract

Instant feedback on sign language can greatly improve sign language interpretation. In this project we plan to use efficient methods to detect the hand correctly using skin detection algorithm and removing all noise using MATLAB and classify the image according to the sign gesture performed. In this paper, we propose an image processing based model which will interpret Bangla Sign Language. The purpose of this model is to find Bangla Sign Language recognition accuracy. The model will detect the skin color of every type using 𝑌𝐶𝐵𝐶𝑅 algorithm and use Bag of features for feature extraction and Support Vector Machine (SVM) for training and evaluation. To validate our proposed model, we used our own dataset of Bangla Sign Languages using hand gestures of both male and female. The average accuracy we got from evaluation set is 86%.