Sign language recognition using SVM

Abstract

There is one of the natural language called sign language. The sign language is for persons who are challenging in hearing and speech, but all people cannot understand this language. So, there is a need of system that can understand what a challenging person wants to say. For this purpose, a system is developed in which sign is captured and processed. The processing includes use of gaussian filter and bicubic interpolation. The work is done in MATLAB. In this system, a person stands in front of camera and makes sign. The captured RGB image is processed. This image is further used to detect edge using canny edge detector and improved sobel edge detector. Improved sobel edge detector uses dilation and erosion operations. The two images produced after edge detectors are fused together. This data is used to extract features. Zone based approach is used to extract features. Further mean and standard deviation are applied on the data in zone based approach. Then Support Vector Machine is used for training of the system.