基于纹理测度与自适应阈值的FCM图像分割算法

New FCM Image Segmentation Based on Texture Measure and Adaptive Threshold

图像分割是模式识别、图像理解、计算机视觉等领域的重要研究内容.基于模糊C均值聚类(FCM)的图像分割是应用较为广泛的方法之一,但其存在抗噪能力差、收敛速度慢等不足.本文以FCM理论为基础,提出一种基于纹理测度与自适应阈值的图像分割算法.该算法首先根据图像局部相关特性,利用Laws纹理测度提取图像特征,并进行图像的FCM初分割;然后结合Otsu准则(最大类间方差法),利用FCM自适应确定阈值,并对初分割结果进行区域合并.仿真实验表明,该图像分割算法的分割结果与人类视觉感知系统具有良好一致性,其不仅能够有效抑制背景噪声,而且提高了图像分割速度.

Image segmentation is a classic inverse problem which consists of achieving a compact region-baseddescription of the image scene by decomposing it into meaningful or spatially coherent regions sharing similarattributes.Fuzzy C-means（FCM）clustering is one of well-known unsupervised clustering techniques,which hasbeen widely used in automated image segmentation.However,when the FCM algorithm is used for imagesegmentation,there are also some problems,such as poor robustness against noise,slow segmentation speed etc.In this paper,we present a novel FCM image segmentation based on texture measure and adaptive threshold.Firstly,the image feature is extracted according to Laws texture measure,and the initial segmentation isperformed on origin image by using FCM algorithm.Then,the adaptive thresholds are computed by utilizing theOtsu rule and FCM algorithm,and a region combination is performed on the initial segmentation image.Experimental results showed the proposed method achieves competitive segmentation results compared to otherFCM-based methods,and is in general faster.