Multi-View Feature Clustering Technique for Detection and Classification of Human Actions

. Recognizing the actions performed by any person is the most suc-cessful applications in pattern recognition. Detecting the action in a moving camera influences dynamic view changes, is based on spatio-temporal information at multiple temporal scales. In this paper, we are presenting a system that is dependent on actions based on multi-view information. These multi-view features are extracted from various temporal scales. The GMM and Prewitt edge filter is used for detecting background and foreground image. The Nearest Mean Classifier is used to cluster features vector’s of moving object. The experiment results demonstrated using Kth dataset producing 98% of accuracy.