Feature Selection Based on AdaBoost in Video Surveillance System

At present, feature-based classification method is widely used in video surveillance system. How to find a group of features which are stable and efficient is concerned by researchers. In this paper, a new method based on AdaBoost is proposed to form a good sub-set of features. This method evaluates the performance of each feature, and then selects features from the extracted features for classification. Under the premise of ensuring the classification accuracy, the speed of the classifier is greatly improved.