Abstract: Potentially misclassified pixels often display when the fuzzy clustering algorithm is used for image segmentation, so a post segmentation method corresponding to refining the segmentation results is proposed. The first step is the detection of segmentation results from the fuzzy clustering algorithms, and then it is natural to find and reclassify the potentially misclassified pixels by evaluating the membership matrix of neighborhood window. The last step is the experimental verification via several test images including both synthetic image and natural images. The results of the experiment demonstrate that the proposed method not only improves the segmentation accuracy of fuzzy clustering algorithm, but also promotes the visual effect of segmentation to some extent.