摘要
针对噪声严重的超声图像,提出了一种结合数学形态学和Level Set的分割方法。首先采用全变差模型进行图像滤波,再通过交互式区域选择和数学形态学方法获得感兴趣目标的二值化图像,并把该二值化图像轮廓作为水平集方法的初始曲线。改进隐式测地活动轮廓模型(GAC)中的边缘检测函数,增强了处理弱边缘的能力。分割结果表明,该方法能够准确地提取出目标轮廓,同时减少了迭代次数和运算时间。
A segmentation method based on mathematical morphology and level set was proposed for ultrasound images. First, the total variation model was utilized to filter the noisy ultrasound image, then the alternate region choosing and mathematical morphology method were used to obtain the binary image of the interesting object. The binary image was used as the initial curve of the level set method. The edge detection function of the implicit geodesic active contour model （GAC） was improved, and the weak edge detection was enhanced. The results show that the target contour can be accurately extracted. While the iterative and computation time is reduced.
出处
《中国测试技术》
CAS
2007年第5期114-117,共4页
CHINA MEASUREMENT & TESTING TECHNOLOGY
关键词
分割
全变差
数学形态学
水平集
Segmentation
Total variation
Mathematical morphology
Level set
