基于直方图和模糊C均值的彩色图像分割方法

摘要

A novel segmentation algorithm of colour image based on the combination of histogram and fuzzy C-means in HSV colour space is proposed in this paper. Firstly the colour image is transformed into HSV space, and considering the singular property of this space, image pixels are divided into singular and non-singular points according to their saturation and intensity. For non-singular points, the 3D HSV colour histogram is built up, and the peaks are screen out using hill-climbing algorithm for pixels FCM clustering; for singular points, the 1D gray histogram is built up, and the peaks are screen out for histogram FCM clustering. Finally, the previous results of segmentation are merged together. Experimental results show that this method can effectively extract colour image of the object and have certain robustness.%提出一种基于HSV空间的直方图和模糊C均值(FCM)相结合的彩色图像分割算法.首先将彩色图像转化到HSV空间,考虑到该空间的奇异性,把图像中的像素点根据饱和度和亮度划分为奇异点和非奇异点,然后对非奇异点建立3D HSV颜色直方图,并用爬山算法筛选出峰值进行像素点FCM聚类,对奇异点则建立1D灰度直方图,筛选出峰值进行直方图FCM聚类,最后合并两种分割结果.实验结果表明,该方法对彩色图像能够有效地提取目标物体,具有一定鲁棒性.