Gait recognition method based on hybrid kernel and optimized parameter SVM

Abstract

The gait recognition algorithm adopt support vector machine based on hybrid kernel function and Parameter Optimization. Partial kernel function and overall kernel function are fitted to compose super-kernel function, so that the SVM obtain better generalization ability and generalization ability. In terms of parameter selection, the text uses the objective function and combine OPS algorithm to select the best kernel parameter. This method makes use of the distance of training samples of different classes to find the optimal (or effective) nuclear parameters instead of the standard SVM training samples. It avoids strong empirical and large amount of calculation of the traditional SVM on model selection. Then the gaits are classified by the support vector machine models. This algorithm is applied to a data-set including thirty individuals. Experimental results demonstrate that the algorithm performs at an encouraging recognition rate and at a relatively lower computational cost.