基于颜色空间和核函数的FCM图像分割算法

摘要

First,we introduce and apply the colour histogram to FCM clustering algorithm to determine the initial number of clusters and initial cluster centres of the fuzzy clustering.Then we bring spatial information to FCM to reconstruct the new membership iteration function containing neighbourhood information.Finally,we replace the original Euclidean distance in original algorithm with kernel-induced distance, and optimise the features of experimental image.The algorithm is also evaluated and compared.Experimental results show:this algorithm hasa good quality and effect in image segmentation,and has a stronger anti-nose ability as well.%首先在确定模糊聚类的初始聚类数和初始聚类中心方面，引入颜色直方图应用于FCM聚类算法中。其次再将空间信息引入到FCM中，重建包含邻域信息的新的隶属度迭代函数。最后，用内核诱导距离取代原算法中的欧式距离，对实验图像的特征进行优化，并对算法进行评价对比。实验结果表明，该算法具有良好的分割质量和效果，并且也具有较强大的噪声抑制能力。