Research For Image Compression Method Based On SVM And Wavelet Transform

With the development of multimedia and communication technology, the people demand higher requirements for the speed and quality of digital image data transmission. Digital image signal contains a huge amount of information, and the limit of channel bandwidth and storage space brought great difficulties for practical applications. Therefore, the compression of image data becomes extremely important. The purpose of image compression is exhaust and transmission the larger original image use as little as possible bytes, and required with good image recovery quality. Based on image compression, image storage and transmission burden will be reduced, so that the high-quality images fast transmission and real-time processing on the Web.Image data compression method has a long research history, but most of the previous work is under the guidance of the traditional information theory. For example, the lossless compression of run-length coding, Huffman coding and arithmetic coding, its compression efficiency by Shannon Entropy restrictions. There are lossy compression, such as forecast coding, DCT transform domain coding, vector quantization encoding which also use only the image signal characteristics of the statistical distribution. Since the late 1980s, due to the establishment of wavelet transform theory, fractal theory, artificial neural network theory and the visual simulation theory, image compression technology got unprecedented development. In recent years, with the further study and development of the Support Vector Machine (referred to as SVM) theory, the research of image compression technology combining with SVM theory also began.Based on the combination of wavelet transform theory and the theory of SVM, this paper takes image compression algorithm as the main research object. This paper introduces the encoding principle, advantages and development prospect of the several existing common image compression algorithm, and highlights the basic principle and algorithm of the image compression standard JPEG2000. It studies the principle of wavelet transform, SVM theory as well as their applications in image processing. on the basis of theoretical research the image compression technology, wavelet transform and SVM, by combining wavelet transform and SVM algorithm, study and realization the wavelet transform and SVM-based image compression algorithm use of MATLAB simulation. Through the study, better image quality has been obtained, and at the same time comparatively analysis has been carried out on the image effect under the different image compression ratio. Experimental results show that, when the Compression ratio is big,, this method still have a very good performance.