核空间自适应非局部均值鲁棒分割算法

为改善传统模糊C均值(FCM)算 法对噪声干扰下图像的分割效果,提出了一种基于核空间邻域信息和自适应非局部均值相结 合的图像鲁棒分割算法。首先,利用传统FCM聚类算法进行初始聚类分割,将 所得聚类中心作为改进算法的初始聚类中心； 其次通过自适应非局部均值算法对图像进行平滑处理,避免图像过度分割；然后对引入核空 间邻域信息的目标函数进行最优化求解, 获取聚类中心和隶属度的迭代表达式；最后对像素进行分类时,利用邻域像素隶属度对像素 进行进一步滤波处理。实验结果表明:本文的 改进算法与传统FCM聚类算法、核空间FCM(KFCM)算法、基于邻域信息的FCM(FCM S)算法和基 于邻域信息KFCM(KFCM-S)算法相比,改进算法对强噪声干扰下的图像具有较好 的鲁棒性,并且能够有效分割图像的目标信 息和背景信息。从几种算法进行聚类分割结果所对应的峰值信噪比(PSNR)也可以看出,本文 改进算法的PSNR更高。

Kernel space adaptive non-local means robust segmentation algorithm

In order to improve the image segmentation effect of the traditional f uzzy C-mean (FCM) algorithm,a robust image segmentation algorithm based on the combination of kernel neighborhood information and adapt i ve non-local mean is proposed.Firstly,the initial clustering segmentation is performed using traditional FCM clustering algorith m,and the clustering center is used as the center of the initial cluster algorit hm.Then,the image is smoothed by the adaptive non-local means algorithm,which avoids the i mage over segmentation.Finally,the objective function for introducing the kernel spatial information is optimally solved,the ite rative clustering center and expression of the degree of membership are obtained ,and finally the pixels are classified and the pixels are further filtered using neighborhood pixels membership.The experimental results show that the improved algorithm under the strong noise interference has go od robustness,and can effectively segment the image information and background information.The peak signal-to-noise ratio (PSNR) of the segmentation results with the proposed algorithm i s also higher.