Key Frame Extraction Using Features Aggregation

In Video Surveillance System, the surveillance of video in its different application such as performing real time online event detection, crime prevention, scene analysis and offline analysis and retrieval of interested events requires very huge computation and memory too. Key frame Extraction (KFE) is selection of frames which represents the object moves and changes in subsequent frames in the video. KFE may be used as pre-processing in surveillance application. Changes in frames are decided on the basis of frame’s features. To describe frame in better way, more than one feature are used. Visual Feature Differences are found which are calculated by considering local threshold and Frame Difference is by considering global threshold. Three features are calculated for finding Frame Difference. Three feature differences are then aggregated using aggregation functions. Different Weights for different features are set according to the contribution of corresponding feature in finding frame difference to get better results. Keywords— Aggregation Function, Feature Extraction, Key frame, Thresholds, Video Summarization.