Application Research Of Spectral Graph Wavelet Transform In Image Compression Coding

At present,the research on the theory and application of multiscale geometric analysis has become a new hot spot.The coding methods based on various types of multiscale geometric analysis have obtained better performance than wavelet encoders to a certain extent.All the methods in wavelet coding have been prefered or improved,such as zerotree,direction tree structure,and important coefficient blocks.However,there is a lack of research of the various transformation coefficients themselves.Therefore,it is of great significance to find a new multiscale geometric analysis method from the distribution law of transform coefficients.The main work of this paper is as follows:(1)Firstly,the image transformation methods applied to image compression are systematically summarized.The theory of multiscale geometric analysis and its application in image coding are introduced in detail.Some representative compression coding methods based on multiscale geometric analysis are analyzed,and their respective advantages,disadvantages and problems are pointed out.(2)Multiscale geometric analysis and sparse representation of spectral graph wavelet are studied.The existing SPECK coding algorithm is improved by combining the characteristics of coefficients,and the advantages of spectral graph wavelet transform of the spectrum are used to improve the accuracy and applicability of the algorithm.which results in more flexible application options of the compression systems.(3)Compression coding method based on spectral graph wavelet,images are first converted to graphs,and the spectral graph wavelet coefficients are obtained by using the spectral graph wavelet transform.The energy of these coefficients is attenuated with the increase of the scales,then the improved SPECK algorithm is used to quantize and encode spectral graph wavelet coefficients.And the original image can be recovered from the sparse coefficients in the case of the image data is compressed.Spectral graph wavelet-based coding method,not only inherited many excellent properties of the classical wavelet transform,but it has the ability to analyze the spectral characteristics of the signal from the perspective of the graph.Experimental results show that the coding method has high efficiency in image compression.Compared with the wavelet transform compression method,PSNR of the reconstructed images is improved and the change is stable.At the same time,a larger compression ratio can be obtained.