Research Of Image Compression Coding Based On Wavelet Transform And Vector Quantization

As one of the main carriers of information in modern society, the apparent feature of the image is an amount of data. There are two problems facing with application as follows: 1) needing huge storage space; 2) requiring high channel capacity while being transported. These problems-solving is crucial to promoting the development of digital image applications. By studying the wavelet transform and vector quantization theory, the present paper proposes an image compression encoding method combining wavelet transform with the improved self-organizing feature mapping algorithm together, which compressed the image data effectively. The main research contents were as follows:The paper begins with the description the background and significance of this paper, as well as the relevant technical research situation in and abroad. Then it introduces the basic theory of image coding, sums up the common image coding method, and also introduces the evaluation method and relevant international standards of image codingIn the second part, it introduces the basic theory of wavelet transform and its application in image compression, which implements the Daubechies9/7 wavelets in image compression application by programming.In the third part, it introduces the basic theory of vector quantization with three commonly used methods of vector quantization design codebook, and compares the function of the algorithm by means of the simulation experiment. It focuses on the introduction to the basic self-organizing feature map (SOFM), and proposes improved algorithms for the weak points as generating the initial codebook, searching winning neuron and amending the codeword weights of winning neuron and topological neighborhood.In the last part, a new image coding scheme is proposed basing on Daubechies9/7 wavelet transform and improved SOFM algorithm. Firstly, the paper transforms the image, then adopts cross-band classification vector strategy to construct a new vector for different features of wavelet sub-band coefficients and uses the improved SOFM to deal with the vector. To verify the effectiveness of the program, the paper combines the wavelet transform with vector quantization together in the pilot experiment, that is to say, compares DWT + LBG, DWT + LVQ and DWT + BSOFM, DWT + ISOFM. The experiment shows that the program enhances to some extent the PSNR values and improves the recovery quality of images.