Image Compression Encoding Algorithm Based On Compressive Sensing And New Feature Fractal

Image compression is an important research topic in information processing.Fractal image compression coding makes full use of image redundancy and self similarity,which can ensure high compression ratio and get better reconstructed image quality.However,baseline fractal compression coding needs to choose matching sub-blocks from a large number of virtual codebook,which causes the long coding time and hinders its application in real life.Aiming at the problem of the long coding time and reconstructed image quality in fractal image compression,the main work of this paper as follows:?1?A fractal compression algorithm based on sum of sub-image cross is proposed to solve the problem of long search matching time and reconstructed image quality.Firstly,the sum of sub-image cross is defined to prove the feature matching relationship between image block matching error and the sum of sub-image cross.Then,combined with the virtual codebook after image threshold classification and dichotomy,the global search is changed to the nearest neighbor search based on the sum of sub-image cross.Compared with the baseline fractal image compression algorithm,the speed of the algorithm in this chapter is increased by 20 times when PSNR is only reduced by 0.93d B.Experimental results show that compared with the basic algorithm,one-norm of normalized block and sum of double cross eigenvalues algorithm,sum of sub-image cross algorithm takes less time when the quality is similar.?2?In order to solve the problem of long coding time,a fractal compression algorithm based on the combination of the sum of sub-image cross and compressed sensing is proposed.Firstly,fractal image compression algorithm based on sum of sub-image cross is applied to the image after wavelet transform,and the difference map is obtained by the difference between the low frequency fractal sub-image and its fractal reconstruction image.Then,the difference map is fused with other high frequency sub-bands,and the optimized l₀ smooth norm compressed sensing reconstruction algorithm is used for coding reconstruction.Finally,reconstructed image is realized by inverse wavelet transform.Compared with the baseline fractal image compression algorithm,the speed of the algorithm in this chapter is increased by 88 times when PSNR is only reduced by 1.67d B.Experimental results show that the algorithm improves the quality of reconstructed image when the time of coding reconstruction is close to other algorithms,and the effect is better than the basic algorithm,sub-block average points algorithm and imitation semi cross trace algorithm.?3?A fractal compression algorithm based on the combination of frame cross ratio feature and compressed sensing is proposed to solve the complex problem of fractal calculation.Firstly,the frame cross ratio feature is defined and then it is combined with the model in the previous chapter.Compared with the baseline fractal image compression algorithm,the speed of the algorithm in this chapter is increased by 84 times when PSNR is only reduced by 1.57d B.Experimental results show that the algorithm improves the quality of reconstructed image when the time of coding reconstruction is close to other algorithms,and the effect is better than the basic algorithm,sub-block average points algorithm and imitation semi cross trace algorithm.Compared with the fractal image compression algorithm based on wavelet transform and the combination model of fractal and the hybrid model of fractal and compressed sensing without using low frequency difference map,the new model is better.