Multi-feature fusion based human action recognition algorithm

A novel hybrid human action detection method based on three descriptors is proposed. Firstly, the minimal 3D space region of human action region is detected by combining frame difference method and VIBE algorithm, and the threedimensional histogram of oriented gradient (HOG3D) is extracted. At the same time, the characteristics of three dimensional global descriptors based on frequency domain filtering (FDF) and the local descriptors based on spatialtemporal interest points (STIP) are extracted. Principal component analysis (PCA) is implemented to reduce the dimension of the gradient histogram and the global descriptor, and bag of words (BOW) model is applied to describe the local descriptor based on STIP to make the video feature dimension consistency. Finally, according to the three characteristics, a linear support vector machine (SVM) is used to create a new decision level fusion classifier, which is used for effective analysis of multi class action. Experimental results show that the proposed feature descriptor has good representation ability and generalization ability. And the proposed scheme obtains very competitive results on the wellknown datasets in terms of mean average precision.