基于有偏场的光栅图像模糊聚类分割算法*

针对有偏场环境下带有光栅的散焦图像分割问题，提出了一种新的基于有偏场估计的模糊聚类分割算法。通过建立依赖于有偏场的模糊聚类目标函数，导出了基于灰度以及邻域灰度均值的聚类中心、模糊聚类函数以及有偏场估计的迭代算法；并在该算法生成的初始分割基础上，利用膨胀算子对分类结果进行细化。该方法较好地处理了传统模糊聚类对有偏场下光栅图像分割精度下降的问题。实验结果表明，基于有偏场的模糊聚类算法能有效分割光栅图像，其分割精度优于传统模糊聚类和阈值法。

A Modified Fuzzy C Means Segmentation Algorithm for Illumination

A novel FCM segmentation algorithm is proposed based on bias field estimation with respect to the segmentation issue of defocused images with illumination patterns under bias field. Firstly the recursive algorithms are provided for bias field estimation, clustering centers and fuzzy clustering function based on intensity and average intensity of pixel neighborhood based on object function of image model with intensity in homogeneity. A dilation operator is then used on initial segmented image for further refinement. The problem of reduced segmentation accuracy with traditional FCM is well solved by the proposed approach. Experiment results show that the proposed method is effectiveness for illumination pattern based images under bias field and of higher accuracy rates than traditional FCM and thresholding.