Detecting human moving shadow by multiple features and SVM

Human target detection and analysis based on shadow biometrics from low altitude airborne platform is becoming a more and more important issue recently. In some scenarios from top view, shadows can provide more details. However, it is the first step to detect the human moving shadows. In this paper, a new framework is proposed to detect human moving shadow by shadow features, which takes SVM (Support Vector Machine) as the classifier, and employs machine learning tactic and random sampling theory to address the above problem. Experiments have shown that this method is an effective and robust approach to detect human moving shadow.