Adaptive Image Coding Algorithm Based On Compressive Sensing

Image and video signal have become one of the most primary information carrier of modern society.In order to accurately reconstruct the original signal from the sampled signal,the traditional sampling process must satisfy the theorem of Nyquist sampling,that is,the signal sampling rate must be greater than or equal to twice bandwidth of the signal.With the increase of people's demand of information such as images and videos,the process of large amount of information including acquisition,compression,storage and transmission,are under great pressure and it has become one bottleneck of the signal and information processing.The emergency of compressive sensing(CS)theory has successfully broken through the limitation of the traditional Nyquist sampling theory,which can simultaneously sample and compress signal.Therefore,it greatly saves the storage space of the signal,which provides a new method to deal with the above problems.In this paper,the theory based on CS has been applied to the coding of images,and the main research results are summarized as follows:(1)An adaptive compression coding scheme is proposed.In the proposed scheme,the proportion of the edge for each image block,according to which different sampling rates are adaptively chosen,is leveraged as the principle to estimate the signal's sparseness degree,so that the high quality of reconstruction is guaranteed at a low sampling rate.Howerver,the ratio of measurements of each block is different when given image's size is huge with too many blocks,which evitably leads to the high cost of video's storage and transmission.Therefore,K-Means method is applied to classify the blocks into many clusters and the same ratio of measurement can be allocated for each cluster.Thus,the quality of the reconstructed image is improved at the same average sampling rate;(2)An adaptive sampling compression scheme based on overlapping block is proposed.In the proposed scheme,the adaptive sampling method is applied to overlapping blocks based on compresive sampling scheme.Considering human visual property,the different sampling rates are allocated to each block according to the percentage of the salient regions.The experimental results show that the proposed scheme achieves a high quality reconstructed image;(3)An adaptive depth image coding scheme based on compressed sensing is proposed.According to the proportion of the edge for each image block,different sampling rates are adaptively chosen for image blocks.Thus,the edge information of the depth image is better protected.Besides,the pixel value of the depth image represents the distance between the object in the scene and the camera,which is used to render virtual view rather than display.Therefore,in order to evaluate the performance of the depth image coding and decoding algorithm,the processed depth image and corresponding color image,which is the "color plus depth" format,are used to sythesize the virtual view image.