Event based video coding architecture

In this work, scalable video codec (SVC) has been used for security surveillance video. The main event in the video is motion of an object. A strategy has been proposed to find out the different motion levels (events) in the video. The level of motion in a Group of Picture (GOP) is used to assign different scalability features to the GOP. The architecture of SVC has been modified to provide this support. The model of the modified SVC architecture is presented in detail. The improved system handles fewer amounts of data for processing and storage yet conveys all the important information related to surveillance video. The implementation and experimental results on the surveillance video has been presented. Results show that the proposed system efficiently detects motion and adapts to the scalability level accordingly.