基于加权L^(1)-L^(2)拟合的模糊活动轮廓图像分割  

Fuzzy Active Contour Model Based on Weighted L^(1)-L^(2) Fitting Energy for Image Segmentation

作　　者：王选 唐利明 WANG Xuan;TANG Liming(School of Mathematics and Statistics,Hubei Minzu University,Enshi 44500,China)

机构地区：[1]湖北民族大学数学与统计学院,湖北恩施445000

出　　处：《湖北民族大学学报（自然科学版）》2023年第4期489-498,共10页Journal of Hubei Minzu University:Natural Science Edition

基　　金：国家自然科学基金项目(62061016,61561019);湖北民族大学科研创新项目(MYK2023040)。

摘　　要：为了对灰度不均匀图像进行分割,结合图像的局部和全局信息,提出基于加权L^(1)-L^(2)拟合能量项的模糊活动轮廓图像分割(fuzzy active contour based on weighted L^(1)-L^(2) fitting energy for image segmentation,FAWFE)模型。首先,利用模糊隶属度函数,基于图像局部和全局信息构建混合模糊拟合图像。然后,构建加权L 1拟合能量项,量化原始图像与混合模糊拟合图像之间的差异,有效地处理灰度不均匀问题。最后,结合L^(2)拟合能量项,保证能量泛函的凸性,使得模型解的存在性和唯一性得以保证,避免陷入局部极小。另外,采用交替迭代算法对FAWFE模型进行数值求解,并与经典的活动轮廓模型对比。结果表明,FAWFE模型不仅可以准确定位目标边界,而且可以在0.6 s左右实现对图像的处理,对于合成图像和真实图像均有良好的处理效果。In order to segment images with intensity inhomogeneity,this paper proposes a fuzzy active contour model based on weightedL^(1)-L^(2) fitting energy for image segmentation(FAWFE)by combining the local and global information of the image.First,a hybrid fuzzy fitting image is constructed based on the local and global information of the image by using the fuzzy membership function.Then,a weighted L^(1) fitting energy term is constructed to quantify the difference between the original image and the hybrid fuzzy fitting image to effectively deal with intensity inhomogeneity.Finally,the L^(2) fitting energy term is combined to ensure the convexity of the energy functional,so that the existence and uniqueness of the solution of the model can be guaranteed,avoiding falling into a local minimum.The FAWFE model is numerically solved using an alternating iterative algorithm.Compared with classic active contour model,the proposed model not only locates the target boundary accurately,but also processes the image in about 1 second,and has a good processing effect on both synthetic images and real images.

关 键 词：图像分割 局部和全局信息 灰度不均匀 混合模糊拟合图像 能量项 交替迭代算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]