Video Compressed Sensing And ARMA Integration Research

Compressed sensing theory and its application is one of the current information science researching hot point in the field of frontier subject, involving many subjects such as mathematics, physics, computer science, optimization, signal estimation, and many other complex problems. Due to video processing has been confusing by problems of dimension in long time, so the compressed sensing was applied in the theory of video processing system to reduce data mount. Aimed at the complexity of the structure of the video signal itself, compressed sensing modeling method based on the conventional excessive, basically from the compressed sensing of image has been used for video frame:a video sequene was treatted as individual such as frame by frame, its advantage is not obvious in the actual video processing well. In this paper, we put forward a new video model which integrates the traditional compressed sening video processing model and Auto Regressive Moving Average (ARMA) model.In this paper, we analyze the problems of the traditional video model and compressed sensing video model, and make innovation in three aspects:the first is that a new video model has been proposed by the further study of the advantages and disadvantages of the traditional compressed sensing video model and ARMA video model; The second is that this new model was given on the theoretical analysis and experiments proving which was feasibility; The third is that noiselet transformation was implemented to reduce noise during video preprocessing state. At the same time, in order to make full use of the structure information of video frame, the compressed sensing of observation model, we use redundant dictionary. In the video reconstruction process, combining with the compressed sensing and linear prediction technology, which static and dynamic parts of video can process respectively. We also use the Model based Cosamp algorithm in reconstruction processing to resolve the/1-/2 mixed norm problem.