基于Contourlet变换的医学图像增强与分割的研究-信号与信息处理专业论文.docxVIP

F 黑龙江大学硕士学位论文Abstract 黑龙江大学硕士学位论文 Abstract WitIl the rapid development of medical imaging technology in clinical diagnosis and treatment，the technology of medical image processing and analyzing has emerged． It call provide the scientific basis for the doctor to assist the diagnosis and treatment． 1he medical image enhancement and segmentation，as the basis and premise for the other medical treatment，have become hot research in the medical image processing field． Tllis thesis has further studied on the eontourlet transfom in the multiscale geometric analysis theory．The contourlet transform Can present the images sparely、Ⅳitll good anisotropy．However,the contour]et transform will cause pseudo-Gibbs phenomena due to the lack of translation invariance．’Then we mainly study the shift-invariant transforill named nonsubsarnpled contourlet transform,which can effectively eliminate the Gibbs phenomenon．Thus，we apply the theory of nonsubsampled contourlet transform to the field of medical image erthaneement and segmentation． nle innovation of this thesis is that the nonsubsampled contourlet transform theory is applied to the medical X-eraniofaeial image enhancement and segmentation algorithm． In the medical image enhancement algorithms based on nonsubsampled contourlet transform，we do the enhancemem simulation experiments诹tll the noisy cephalogram images after we focused on the characteristics of medical X-my film．Experimental results demonstrate that the proposed algorithm outperforms histogram equalization， wavelet—based and contourlet·based enhancement algorithms in terms of human vision． Not only the contrast is more distinct and meanwhile the noise is restrained，but also the details of the image edge are enhanced and the values of PSNR ale increased．Then we apply the model method of the discrete fractional Brownian motion to do the ．¨． 「 「 segmentation segmentation simulation experiments with the enhanced cephalogram images· Experimental results demonstrate t