摘要
基于区域信息的主动轮廓模型应用在图像分割中,难以使初始轮廓线的鲁棒性和分割强度异质图像的能力实现有效统一。针对这一缺陷,根据Fisher判别准则,在基于全局区域信息和局部区域信息的主动轮廓模型的基础上,对局部区域信息项进行了变换以及引入了判别准则,得到了一种改进的基于区域信息的主动轮廓模型。改进的模型不仅增强了初始位置的鲁棒性,而且可以有效处理强度异质图像,通过实验检验了该模型的性能。
In image segmentation, active contour model based on regional information application is difficult to make the robustness of initial contour line and the ability of segmenting heterogeneous image segmentation unified effectively. In view of these disadvantages, according to the principle of Fisher discriminant criterion, this paper gets an improved active contour model based on regional information, which is on the basis of active contour model with the global regional information and local regional information, by transforming local regional information item and introducing the discrimination criterion information. The improved model not only enhances the initial position of robustness, but also can effectively deal with intensity heterogeneous image. The property of this model is verified by experiments.
出处
《红外技术》
CSCD
北大核心
2016年第4期325-332,共8页
Infrared Technology
基金
广西高校科研项目(KY2015ZD127)
湖南文理学院博士科研启动基金项目
广西高校科学技术研究项目重点项目(ZD2014129)
关键词
主动轮廓模型
图像分割
判别准则
active contour model
image segmentation
discrimination criterion
