Improvement Of Compressed Sensing Algorithm And Its Application In Image And Video Coding

The image and video signal is the most important information carrier in the contemporary life. Compressed sensing theory has broken the Nyquist criterion which can sample the signal at the same time compress the signal and save the storage space. In this paper, we focus on the compressed sensing reconstruction algorithm and its application in image and video coding, we made thorough research works in the following three aspects.First of all, there are several questions of the Total Variation algorithm such as blurred image and excessive number of iterations which use the random sampling of the time-domain signal as the initial iterative value. The reconstruction algorithm was improved as follows:We take the inverse transform of the frequency domain signal as initial iterative value. The measurement matrix was designed to capture more low frequency coefficients. Experiments show that our method can improve the quality of the reconstructed image.Then, there are some questions with the image coding method which was based compression sensing theory such as disable to adaptive choosing coding methods. So we propose an adaptive selection algorithm which was based on edge information of image. For the edges of the image is basically the same with the image gradient, the image blocks was coded with compressed sensing theory which contains more image edge information. The others were coded with the traditional methods which contain few edge information. So we get the goal that the image block can adaptively choose the coding methods. Experiments show that our adaptively choosing coding method can improve the quality of the reconstructed image.Finally, there also exits shortcomings in the inter-frame coding of video. The adaptive coding method of residual image was proposed to solve these questions. The first frame of the GOP was coded as the intra-frame with adaptive coding method based on compressive sensing. The motion estimation was introduced into the methods which were to obtain the residual image. We use the adaptive coding method to code the residual image. Experiments show that our method can improve the quality of the reconstructed video.