Recognition of inter-class variation of human actions in sports video

Human action recognition encompasses a scope for an automatic analysis of current events from video and has varied applications in multi-various fields. Recognizing understanding human actions videos still remains difficult downside as result the massive variations look, posture body size inside identical category. This paper focuses on specific issue related to inter-class variation Action Recognition. To discriminate among category, novel approach which is based wavelet packet transformation feature extraction. As we are concentrating classifying similar non-linearity features analyzed discriminated by proposed Deterministic Normalized –Linear Discriminant Analysis (DN-LDA). However major part system relays classification dynamic feeds classified Hidden Markov Model at final stage rule set. With trained dataset rules framed with end user, our intelligent HAR capable achieving accuracy rate 97.4% higher than other state art approaches. Experiments results have shown that discriminative well adapted variation.