Support Vector Machines (SVM) is a kind of novel machine learning methods which has become the hotspot of machine learning because of excellent learning performance. It also has successful applied in some fields which are almost pattern recognition, such as face detection, handwriting digit recognition, text categorization, etc. But as a new technique, SVM also has many application fields that need to be researched. In this thesis, Image processing has been researched using the theory of Support Vector Regression based on this researching environment. The corresponding working are given as follows:1.The algorithm for Support Vector Machines and Statistical Learning Theory is introduced then the method of image representation by Support Vector Regression is proposed.2.Traditional methods of image processing are applied to SVR image which is provided a base for research on SVR image, for example, image geometrical changing, image Gaussian filter, image enhancement, SVR image compression and reconstruction.3.The image intensity of neighborhood of every pixel is well estimated by Least Squares Support Vector Machines(LS-SVM) and the gradient operators and corresponding zero crossings operators are obtained by LS-SVM. Then the method of edge detection based on the combination result of the gradient and zero crossings is proposed. Experiment results of edge detection by LS-SVM are satisfied.4.The filter operator is deduced by Least Squares Support Vector Machines in wavelet image based on the principle of image denoising by wavelet soft-thresholding. Then the processing of image denoising using the operators is given and the experimentsresults show that the proposed denoising technique is effective in removing Gaussian noise and preserving edge information well. |