M6. Video Analysis

The goal of this project is to learn the basic concepts and techniques related to video sequences processing, mainly for surveillance applications. We will focus on video sequences from outdoor scenarios, with the application of traffic monitoring in mind. The main techniques of video processing will be applied in the context of video surveillance: moving object segmentation, motion estimation and compensation and video object tracking are basic components of many video processing systems. In a first stage, moving object segmentation will be tackled considering scenarios with static camera. Afterwards, camera motion will be considered. Tracking of the moving objects can be performed in both scenarios. The tracking result provides high level information that can be analysed for traffic monitoring. The learning objectives for the students are the use of pixel based statistical models (such as mixture of gaussians) for modeling a scene background and for moving object segmentation, the development of optical flow estimation methods for camera motion compensation, and techniques for object tracking (ranging from simple blob analysis to more complex techniques based on filtering and probabilistic data association). The performance of the developed techniques will be measured using standard metrics for video analysis.