A Video Motion Object Segmentation Research Based On Cellular Neural Network (CNN)

Video object segmentation can serve for much application in computer vision area. Such as video coding, video authoring and edit, video search, video-based monitoring, etc. Video segmentation is a kind of technology which is applied widely. One of the important questions, which video segmentation is faced with, is difficult to satisfy real time require, because too much data need to operate. The characteristic that CNN is fit for image progress and the speed of its chip is high can give great help to solve this question. Now video data become larger and larger, so the real time is required higher and higher. Using the arithmetic and hardware related with CNN is an effective method to solve these questions. The video segmentation algorithm research based on CNN has broad application foreground. Now some American and European research government are taking up with the video segmentation based on CNN, which is new way and the foreland of neural network application field.In this paper, the research motive of cellular neural networks application in video motion segmentation is discussed. After reviewed and compared the existing video segmentation methods, the previous application of CNN in that field is introduced. The quantization method of CNN input and output in image processing and the basic principle of CNN applied to image processing is also introduced. Aim at video sequences with static background, a video motion object segmentation algorithm based on CNN is presented. The algorithm is realized on Matlab7.0 and its experimental results on video sequences confirm that the proposed algorithm is feasible.