Human Activity Recognition Using Gabor Wavelet Transform and Ridgelet Transform

A methodology for the recognition of human activity in a video sequence based on Gabor wavelet transform (GWT) and Ridgelet transform (RT) is presented in the paper. The scale and rotation invariant property of GWT and orientation dependent property of RT is used to determine the feature space that represents a particular human activity. GWT and RT are applied on the key pose of activity and both techniques offer the analysis of image sequences at different level of resolution. The proposed methodology is validated on the challenging publicly available dataset and the accuracy of the method is calculated in terms of average recognition rate (ARR). The performance of the methodology is compared with the similar state-of-the-art and gives the improved performance in comparison with the others technique.