Relational Match with Parallel Interpretation Tree for Recognition of Part-Based Human Body

In recognition of part-based human body, relational match is to find a coincidence mapping from object body parts to scene parts using the interpretation tree searching. The number of interpretation tree routes can be increased exponentially in practical application. In this paper, the interpretation tree was built between layers in parallel by divided object body part-sets, and a parallel mechanism between branches was used to search the interpretation tree. This paper described the principle, process and implement method of the relational match with parallel interpretation tree for recognition of part-based human body. The experiment results showed the effectiveness of the proposed approach.