A Noval Approach to Automated Video Analysis using DECOLOR Algorithm

Video analysis is very important for many applications such as video surveillance, traffic analysis, vehicle navigation etc. Object detection is fundamental step in video analysis. Object detection is the process of finding instances of real world objects such as faces, bicycle, and building in images or videos. It is challenging to support moving object detection in timely manner. Moreover to automate the video analysis, object detection requires training on some labeled examples. Without this training it is very difficult to analyze moving object. Some methods have been proposed to tackle this problem and these are based on motion information; but these methods are not useful when there is non-rigid motion or dynamic background. These challenges can be addressed in framework called as Detecting Contagious Outlier in the Low-rank representation. It is assumed in DECOLOR that the underlying background images are linearly correlated and the low rank matrix is formed from vectorized video. The moving objects can be detected as an outlier in low-rank representation. This method also integrates the object detection and background learning into single process.