基于改进边缘活动轮廓模型的超声图像分割  

Ultrasound image segmentation based on improved edge active contour model

作　　者：倪晓航 肖明波[1] NI Xiao-hang, XIAO Ming-bo(School of Communication Engineering, Hangzhou Dianzi University, Hangzhou 310018, Chin)

机构地区：[1]杭州电子科技大学通信工程学院,浙江杭州310018

出　　处：《计算机工程与设计》2018年第6期1675-1678,1749,共5页Computer Engineering and Design

摘　　要：为克服传统边缘函数(edge-stop functions,ESFs)不能停止模糊边界问题,提出一种方法构建一组边缘停止函数的活动轮廓模型实现模糊边界的分割。该方法包括一组带有灰度信息和概率评分的标准分类器,ESF可以由任意分类算法构造,将其应用到基于边缘的水平集分割方法中。采用距离正则化水平集演化方法结合k-近邻算法(k-nearest neighbor,kNN)或支持向量机(support vector machine,SVM)对超声图像进行分割,实验结果表明,该方法能够有效分割超声图像,明显优于其它分割方法。To overcome the problem that the traditional edge-stop functions（ESFs）fail to stop the poor boundary contours,a framework was proposed,which constructed an active contour model of a group edge function to realize the segmentation of fuzzy boundary.The framework included a set of standard classifiers with gray information as well as probability scores.ESFs could be constructed using any classification algorithms and applied to any edge-based model.Experiments on ultrasonic image were carried out using distance regularization level set for edge-based active contour models as well as k-nearest neighbor algorithm（kNN）or support vector machine（SVM）.Experimental results show that the proposed method can effectively segment the ultrasound image,which is superior to other segmentation methods.

关 键 词：基于边缘的活动轮廓 边缘停止函数 梯度信息 超声图像分割 概率得分 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]