A Grey Wolf Intelligence based Recognition of Human-Action in Low Resolution Videos with Minimal Processing Time

The usage of video cameras for security purposes has grown in recent years. The time for recognition of human plays an important role in solving many real time problems. In this paper, the process for identifying human action is done by separating the background using local binary pattern (LBP) and features extracted using faster histogram of gradients (FHOG) and Eigen values based on power method. The features are combined and optimized using grey wolf optimization (GWO) and finally classified using support vector machine (SVM). The experimental results are compared with existing methods in identifying the human action. The time factor is evaluated and compared with different optimization techniques like particle swarm optimization (PSO), Firefly algorithm (FA) and grey wolf optimization. The entire process is performed on three well known datasets like VIRAT dataset, KTH dataset and Soccer dataset. The comparison results prove that the recognition is done in quick time i.e. 10.28sec with improved rate of accuracy 93.35% for soccer dataset using proposed method.