Automatic Key Posture Selection for Human Behavior Analysis

摘要: A novel human posture analysis framework that can perform automatic key selection and template matching for behavior is proposed. The entropy measurement, which commonly adopted as an important feature to describe the degree of disorder in thermodynamics, used underlying identifying postures. First, we use cumulative change indicator select appropriate set postures from a video sequence then conduct cross check remove redundant With detected stored templates, similarity between query database evaluated using modified Hausdorff distance measure. experiment results show proposed system highly efficient powerful