Gait Recognition Based on Feature Fussion and Support Vector Machine

Facing the problem that the recognition rate based on single feature algorithm is low in present gait recognition studying, a new gait recognition algorithm based on feature fusion and support vector machine is proposed. First, getting the body contour image by using background subtraction, then extracting the human body contour line and using Fourier descriptor to express human body contour feature, extracting limb angles feature, also introducing a kind of dynamic regional variance feature. Fusing them to get a combination features vector and finally using support vector machine to classify the vector. A plenty of experiments are made on the CASIA gait databases. Experimental results show the high recognition rate of this method.