一种基于平衡多小波分析与SOFM的图像分割算法研究  

Algorithm of Image Segmentation Based on Balanced Multi-wavelet Analysis and SOFM

作　　者：李鑫环[1] 陈立潮[1] 张英俊[1] 赵勇[1] 

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024

出　　处：《计算机与数字工程》2009年第11期155-158,185,共5页Computer & Digital Engineering

摘　　要：针对目前MRI脑图像分割算法在图像分割速度和精度上不理想的问题,提出了一种将平衡多小波分析与SOFM相结合的BMSOFM算法。该算法在对MRI脑图像进行平衡多小波分析的基础上,用SOFM对图像聚类,最终得到分割结果。对不同分辨率的MRI脑图像的仿真实验表明,在高分辨率的情况下,BMSOFM不但加快了分割的速度,而且提高了聚类精确度,分割效果得到明显改善。In allusion to the problem that MRI brain image segmentation is not so perfect at the speed and precision at present, a method based on the balanced multi-wavelet and SOFM （BMSOFM） is proposed in this paper. This algorithm based on the balanced multi-wavelet transform to the MRI images, uses SOFM to cluster the image, finally the segmented image is got. MRI brain images of different resolution are used in the experiment, and are segmented by the method of BMSOFM and the traditional SOFM. Compared the results, it is indicated that in high resolution BMSOFM not only mends the speed of the segmentation, but also enhances the accuracy in clustering, and excels the traditional SOFM at the result.

关 键 词：图像分割 平衡多小波分析 自组织特征映射 核磁共振图像 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]