Improved Canny Edges Using Cellular Based Particle Swarm Optimization Technique for Tamil Sign Digital Images

The development of computer based sign language recognition system, for enabling communication with hearing impaired people, is an important research area that faces different challenges in the pre-processing stage of image processing, particularly in boundary detection stage. In edge detection, the possibility of achieving high quality images significantly depends on the fitting threshold values, which are generally selected using canny method, and these threshold values may vary, based on the type of images and the applications chosen. This research work presents a novel idea of establishing a hybrid particle swarm optimization algorithm, which is a combination of PSO with the behavioural pattern of cellular organism in canny method, that defines an objective to find optimal threshold values for the implementation of double thresholding hysteresis method, which is viewed as a non-linear complex problem. The attempt to incorporate the model has minimized the problem of quick convergence of PSO algorithm which has improved the detection of broken edges. The efficiency of the proposed algorithm is proved through the experimental observation, done in Tamil sign images to indicate the better performance of canny operator by introducing new variant based PSO.