Object Tracking via Boosted Cascade of Simple Features and Coarse and Fine Structural Local Sparse Appearance Models

Sparse coding technique is effectively implemented in visual tracking applications to find the position of target in different frames. A large number of methods will choose holistic appearance of the target.it is more risky under occlusion. In this project, we develop a robust tracking algorithm based Haar features and Adaboost classifier for fast detection and a coarse and fine structural local sparse appearance model for the case of occlusion. The proposed method creates a dictionary for a target by using some templates. The averaging and pooling operations extract consistent appearance of object parts, thereby helping for tracking the target accurately. This paper compare the accuracy and running time of sparse coding based occlusion detection model and the proposed work to highlights the performance of the proposed frame work.