摘要
针对模糊C均值聚类(FCM)算法存在对噪声和图像中的异常值非常敏感的缺点,本文提出了一种基于空间信息的FCM图像分割算法(Fuzzy spatial information C-means clustering,FSICM)。该算法有效的利用图像中的空间信息和待分割图像的灰度信息,提高了对噪声图像的分割精度。首先将得到的空间函数引入到隶属度函数中改变了每个像素点的隶属度权重,以此来抑制噪声。然后,将图像中像素的邻域信息考虑进来,对每个像素进行处理形成线性加权和图像,对新形成的图像的灰度直方图进行聚类。通过对合成图像和真实图像的实验结果表明,FSICM算法对噪声图像的分割精度均在99%以上,对比其他算法,本文提出的算法能够在保持图像重要细节的同时去除噪声,对噪声图像具有鲁棒性。
In this paper,we present a fuzzy spatial information c-means clustering(FSICM)algorithm for image segmentation by introducing spatial information,in order to overcome the defects of being sensitive to noises and outliers in image brought by Fuzzy c-means clustering(FCM)algorithm.The algorithm effectively utilizes the spatial information of the image and the gray information of the image to be segmented,and improves the segmentation accuracy of the noisy image.Firstly,the spatial function is introduced into the membership function to change the membership weight of each pixel so as to suppress the noise.Then,the neighborhood information of pixels in the image is taken into account,each pixel is processed to form a linear weighted image,and the gray histogram of the newly formed image is clustered.Experimental results of both synthetic and real images show that the segmentation accuracy of FSICM algorithm for noise image is above 99%.Compared with other algorithms,the algorithm proposed in this paper can remove noise while maintaining important details of the image and is robust to noise images.
作者
朱素霞
祖宏亮
孙广路
ZHU Su-xia;ZU Hong-liang;SUN Guang-lu(School of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, China)
出处
《哈尔滨理工大学学报》
CAS
北大核心
2020年第4期101-108,共8页
Journal of Harbin University of Science and Technology
基金
国家自然科学基金(61502123)
黑龙江省青年科学基金(QC2015084)
中国博士后科学基金(2015M571429).
