Human Behavior Recognition: An l_1 - l_s KSVD-Based Dictionary Learning and Collaborative Representation-Based Classification

摘要

This work presents a new idea for human behavior recognition based on dictionary learning algorithm and collaborative representation-based classification approach. In this paper, we have proposed an l_1 - l_s-based KSVD algorithm for learning a dictionary and collaborative representation is used in the classification phase for this problem. The performance of our proposed idea for human behavior recognition problem establishes the superiority of our new idea.