Human Motion Retrieval Based on Feedback Learning

A novel content-based feedback learning motion retrieval approach is presented. The approach includes two primary stages. (1) In the learning stage, fuzzy clustering is first utilized to get the representative frames of motions, and the body gesture features are extracted to build a motion feature database. (2) In the motion retrieval stage, the query motion feature is extracted according to stage (1). Similarity measurements are then conducted by employing a novel method that combines Manhattan distance dynamic programming (MDP) and support vector machine (SVM) feedback learning. The retrieval results are ranked according to the feedback learning values. The effectiveness of the proposed method is verified experimentally.