基于改进LGDF模型的超声图像自动分割方法

删除或更新信息，请邮件至freekaoyan#163.com(#换成@)

本站小编 Free考研考试/2024-01-16

-->

朱永杰,邱天爽.基于改进LGDF模型的超声图像自动分割方法[J].,2016,56(1):28-34

基于改进LGDF模型的超声图像自动分割方法

Automated segmentation method for ultrasound image based on improved LGDF model

DOI：10.7511/dllgxb201601005

中文关键词:局部熵超声图像自动分割局部高斯分布拟合能量(LGDF)正则化项

英文关键词:local entropy ultrasound image automated segmentation local Gaussian distribution fitting energy (LGDF)regularized term

基金项目:国家自然科学基金资助项目(81241059,61172108)；“十二五”国家科技支撑计划资助项目(2012BAJ18B06-04).

作者	单位
朱永杰,邱天爽

摘要点击次数:1462

全文下载次数:1494

中文摘要:

基于局部高斯分布拟合能量(LGDF)模型的图像分割方法，对初始轮廓选取及参数选择较敏感．如果初始轮廓手动选取不当会由于陷入局部极小值而导致分割失败，且分割速度较慢．针对以上不足，提出了一种改进的LGDF模型的超声图像自动分割方法．该方法的正则化项由具有双极值点的势函数构成，在水平集函数进化过程中，可以避免由单极值点势函数造成的水平集函数震荡和扭曲，从而加快了收敛；另外，将局部熵阈值分割的结果作为LGDF模型的初始轮廓，接近真实轮廓，可以克服手动选取初始轮廓的影响．实验结果表明，该方法能自动获取合适的超声图像初始轮廓，并得到较好的分割结果，同时大大提高了分割速度．

英文摘要:

The image segmentation method based on local Gaussian distribution fitting energy (LGDF) model is sensitive to initial contour and parameter selection. If initial contour chosen manually is not suitable, the segmentation will even fail because of being lost in local minima. In addition, the segmentation speed is slow. To solve these problems, an ultrasound image automated segmentation method based on improved LGDF model is proposed. This method′s regularized term formed by double-poles potential function can avoid the oscillation and distortion of level set function caused by single-pole potential function in the process of level set function evolution, which accelerates convergence. Besides, the result of local entropy threshold segmentation is regarded as the initial contour of LGDF model and close to the true contour, which overcomes the impact of manually selecting initial contour. Experimental results show that this method can automatically obtain suitable ultrasound image initial contour and get preferable segmentation result. Meanwhile, the speed of segmentation is greatly improved.

查看全文查看/发表评论下载PDF阅读器

关闭

基于改进LGDF模型的超声图像自动分割方法

本站小编 Free考研考试/2024-01-16

相关话题/

领限时大额优惠券,享本站正版考研考试资料!