The Study Of Improved Medical Image Enhancement Based On Beyond Wavelets Transform

This paper analyzes the beyond wavelets algorithm appeared at home and abroad and proposes an improved image enhancement algorithm based on their shortcomings. Considering that the image cannot be approximated by the wavelet transform ideally. This paper introduces the contourlet transform instead of wavelet transform which make the coefficients sparser. In the process of denoising, most of the algorithms focus on improving the structure of threshold and threshold function, but ignore the re-processing of denoising results. For improving this, we improve the threshold and threshold function and extend it to the contourlet frame. Then we take the mean image of spinning denoising images, which make the image distortion improve to some extent. In the enhancement process, most of the algorithm uses the global gain function. For improving this, this paper designed two relevant gain functions for the low-frequency sub-image and the high-frequency sub-images. The improved algorithm can remove more noise and make the texture clear er. Using Matlab to carry out the simulation experiments, we enhance a mammography X-ray image by six algorithms. The six algorithms are histogram modification met hod 、 Laplaces equation、original wavelet algorithm、improved wavelet algorithm、original contourlet algorithm and improved wavelet algorithm which proposed in this paper. The results show that the proposed algorithm is the best. Compared with other results, t he dealt image is brighter in general and the details of the image are clearer. The calcifications and masses in the breast are more easily to observe and the textures are clearer, no large white.Tetrolte transform, a new beyond wavelet, can be used to represent images sparsely. The tetrolet transform has a strong efficiency for image approximation. However, the original image denoising algorithm based on tetrolet transform make blocking artifacts appearing. In order to reduce the blocking artifacts, recursive cycle spinning is introduced which can reduce the fuzzy block. To avoid the recursive cycle spinning changing too many coefficients to zero, a suitable window threshold function is constructed. Using Matlab to carry out the simulation experiments, we denoise a MIR image. Compared with improved contourlet algorithm and original tetrolet algorithm, our proposed algorithm makes the best result. The proposed algorithm gives better performance in PS NR,removes more noise and preserves more significant information of local features.