Research On Image Segmentation Algorithm Based On Fuzzy C-Means Clustering

Image as a carrier for human to know and understand the world is an important way to express, analyze and transmit information. Image processing by using computer technology is getting more and more attention. Image segmentation plays a key role in the fields of image processing. It is one of the most important challenging problems in image processing. Fuzzy clustering is one of the most commonly used methods and has been successfully applied in image segmentation. It has been widely used in the fields of medical image processing, object identifying, etc.Among the fuzzy clustering algorithms, fuzzy c-means(FCM) algorithm is the most popular algorithm used in image segmentation, since it has robust characteristics for ambiguity and can retain more information of the original image. However, the standard FCM algorithm does not consider any spatial information which can’t make it achieve good segmentation results in low contrast, in-homogeneity and noisy images. In order to overcome the problems mentioned above, many modified FCM algorithms with spatial information incorporated have been proposed. The modified FCM algorithms incorporate spatial information mainly through modifying the objective function or modifying the similarity measurement between pixels and cluster centers. Although the modified FCM algorithms have improved the performance of standard FCM algorithm to some extent, they still have the following disadvantages:(1) lacking enough robustness to noise and outliers;(2) needing to execute a lot of iterations to achieve convergence resulting in low efficiency. This paper focuses on improving the performance of FCM algorithm by fully and effectively using the spatial information from different angles. The main research work and achievements are as follows:(1) This paper studies FCM algorithm and other improved FCM algorithms including FCM_S algorithm with spatial information incorporated, SAFCM algorithm based on spatial information, etc.(2) This paper proposes a fast and robust FCM algorithm for image segmentation. Firstly this algorithm modified the objective function of FCM algorithm by redefining the distance between the pixel and cluster. Then it constructed a spatial function by combining pixel gray value similarity and membership to update the membership in each iteration. During each iteration, FRFCM algorithm calculates each pixel’s membership based on the objective function incorporating spatial function of FCM_S algorithm and then it updates the membership using the newly-constructed spatial function. By comparing the segmentation results of different images of FRFCM algorithm and other modified FCM algorithms, it can be concluded that FRFCM algorithm can achieve ideal result in less iteration, providing great robustness to noise.(3) This paper also proposes an improved anti-noise FCM algorithm for image segmentation. This algorithm introduces a new neighborhood term to the standard FCM algorithm by combining the relative location information and the gray level information of the neighboring pixels. The qualitative and quantitative experiments results for images with different noisy levels illustrate that the proposed algorithm not only improves the robustness to noise, but also can identify objects from images whose gray level difference is small.