基于特征散度的自适应FCM图像分割算法

An Adaptive FCM Image Segmentation Algorithm Based on the Feature Divergence

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,但其存在距离测度鲁棒性差、需预先给出初始聚类数目、未考虑图像局部相关特性等问题。为克服上述缺点,通过引入特征散度进行距离测度,并结合聚类有效性指数自适应确定初始聚类数目和根据Laws纹理测度提取图像特征等措施,提出了一种新的FCM图像分割算法。实验结果表明,该新算法可以有效地提高图像的分割效果(特别是纹理图像),其分割结果优于现有FCM图像分割方案。

Fuzzy C-means （FCM） clustering is one of well-known unsupervised clustering techniques, which has been widely used in automated image segmentation. However, when the classical FCM algorithm is used for image segmentation, there are also some problems, such as weak robustness of distance measure, reguire-ments of setting the initial number of clusters in advance, without considering local image feature. In this paper, an adaptive FCM image segmentation algorithm based on the feature divergence is proposed, which can accomplish image segmentation by importing the feature divergence vector into distance measure, incorporating the cluster validity exponent to ascertain the initial number of clusters automatically and extracting the image feature according to Laws texture measure. Experimental results show that the proposed method is simple and work well for most images （especially for texture images）, and has better segmentation effect than the existing FCM image segmentation.