乳腺图像的非刚体特性使其不能通过刚体的匹配方法进行匹配。为此，提出一种基于分层算法的肿块匹配方法。利用最大类间方差法进行阈值分割以截取胸肌区域，使用最小二乘法进行胸壁线的拟合，找出乳头以及中轴线的位置，建立局部坐标系确定匹配条形区域带。采用分层算法在条形区域带内进行疑似病灶区域的提取，通过加权互信息相似性度量实现肿块匹配。对分层算法肿块匹配方法进行可行性分析和实验验证。选取100对图像进行肿块匹配实验，结果表明，相对于非分层算法，分层算法的匹配结果更有效，匹配精度达到86%。%The breast image cannot match by using the method of rigid matching because of its non-rigid characteristics. To solve this problem, this paper proposes a mass matching method based on layering algorithm. Use the method of Otsu threshold segmentation to intercept the pectoral region and use the least-squares method to fit the chest wall line, and establish local coordinate system to identify the matching bar area by finding out the position of nipple and central axis. It extracts the suspected lesion area in the bar areas by using layering algorithm, and by measuring the similarity based on the weighted mutual information to realize the mass matching. It analyzes the feasibility of the stratified mass matching method and does experiments. This paper selects 100 pairs of images to carry out the experiments of mass match. The experimental results show that the layering algorithm for mass matching is more effective than the non-layered algorithm, and the matching precision reaches 86%.
展开▼
