基于最大类间方差的权重自适应活动轮廓模型  

Weight-self adjustment active contour model based on method of maximum classes square error

作　　者：赵怡 邓红霞[1] 张玲[1] 李钢[1] ZHAO Yi;DENG Hong-xia;ZHANG Ling;LI Gang(College of Computer Science and Technology,Taiyuan University of Technology,Taiyuan 030024,China)

机构地区：[1]太原理工大学计算机科学与技术学院,山西太原030024

出　　处：《计算机工程与设计》2018年第2期486-491,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61373101;61472270;61402318);山西省自然科学(青年科技研究)基金项目(2014021022-5)

摘　　要：针对传统的活动轮廓模型不能有效处理灰度非均匀图像以及权重参数需要手动调节等问题,提出一种基于最大类间方差的权重参数自动调节模型。将最大类间方差的思想引入LIF模型并作为局部项,通过最大化目标和背景的类间差异提高分割准确度;将C-V模型作为全局项降低对初始轮廓的敏感性;引入图像熵自适应调节局部项与全局项的比重,提高运算效率和精度。实验结果表明,该模型对多种灰度非均匀和噪声图像均可实现理想分割,对初始轮廓的位置不敏感,提高了分割效率。In view of the existing problems in the traditional active contour model that it cannot deal with images with intensity inhomogeneity effectively and that the weight parameter need to be selected manually,an active contour model with weight-self adjustment based on the method of maximum classes square error was proposed.The method of the maximum classes square error was integrated with the LIF model as the local term.Thus the segmentation accuracy was improved though maximizing the difference between the target area and background area.The sensitivity to the initial contour was reduced by the global term of the C-V model.The image entropy was introduced to adaptively adjust these two terms which improved the segmentation efficiency and accuracy.Experimental results show that the proposed model achieves better performance on handling a variety of images with noise and intensity inhomogeneity and it is less sensitive to the position of initial contour.The segmentation efficiency is improved.

关 键 词：最大类间方差 LIF模型 图像熵 活动轮廓模型 图像分割 水平集 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]