Compressed Sensing Image Reconstruction Algorithm And Distributed Video Compression Perception System Research

Compressed sensing theory which emerged in recent years breaks through thelimitations of the Nyquist sampling theorem when used to acquire Signal and it pointsout that as long as the signal in a certain transform domain is sparse, can reconstruct theoriginal signal by solving an optimization problem using a small number of randomlinear projection of this signal. This feature greatly simplifies the process of signal datacompression and has broad application prospects in the video image signal processing,radar imaging, medical imaging, wireless sensor networks, etc.Traditionally, interframe joint coding is generally taken in under the situation oflow-power video coding which need encoder which has powerful computation abilityand can low-power requirements are often difficult to achieved.Distributed Video Coding has caused extensive attention in recent years and it hasdifferent coding strategies compared with the previous method. Coder independentlycodes in the coding end of each frame in the video sequence and then the decoderdecodes according to the high correlation between adjacent inter-frame in time domain.This encoding strategy shifts the large amount of calculation of the inter prediction tothe decoding side and makes the encoder complexity lower, power control simpler andit is in line with the scenarios of low-power.Compressed sensing image reconstruction algorithms and distributed video codingbased on compressed sensing is researched in this paper.This paper firstly reviews and summarizes some of the existing algorithms of theimage reconstruction and focuses on the model of compressed sensing theory, andpresents a new structure model based on the edge of the texture image reconstructionalgorithm which is based on the high frequency sub-band image wavelet coefficients inthe image signal edge of the texture part of the cluster distribution characteristics ofcompressed sensing. The algorithms fully tap the image wavelet sparse distributioncharacteristics of any high frequency sub-band coefficient, making the imagereconstruction accuracy improved. The experimental results show that the proposedalgorithm is superior to standard CoSaMP and Tree-based CoSaMP. In addition, this paper also proposes a new distributed video coding scheme basedon compressed sensing and the encoder is independently coded and the decoder decodesaccording to interframe information from the measurement data transmitted from theencoder. In the decoding process, in order to make better use of the interframeinformation, we use the interframe interpolation method to generate the side information.Meanwhile we can achieve the secondary information of prediction residual forreconstruction and the decoder can achieve higher quality of the video frame forreconstruction according to the structural characteristics of compressed sensingreconstruction algorithm and the prediction residual. The experimental results show thatthe framework does well in the video compression and reconstruction.