Remote-sensing Image Segmentation Based On Fuzzy C-means Algorithms Theory

Remote-sensing image segmentation algorithm is a crucial step in the image processes and analysis. Remote-sensing image segmentation is the process and technology that to divide images into the regions with distinctive features and extract the target interested in. Remote-sensing image with characteristics of multiple gray level, more informative, fuzzy boundary, complex target structure, influence of the noise and so on. Which make it ask for higher requirements to remote-sensing image segmentation whether in the efficiency or the effect, but lack of a reliable model to guide it completely, it block the application of segmentation technology in the field of remote sensing to some extent. To the current, there has not been an ordinary and objective image segmentation algorithm which can enable various types of images achieve optimal segmentation effect.Recently, kinds of segmentation methods were proposed, These algorithms are based on the different principles. This paper proposes some modified remote-sensing image segmentation methods based on the study of fuzzy c-means algorithms and threshold-based segmentation methods, specific tasks are as follows:1,The conventionally standard FCM algorithm is sensitive to noise and intensity inhomogeneity. An improved FCM based algorithm is proposed in this paper, which firstly models the noises of image as a slowly varying additive or multiplicative noise and iteratively approximates the intensity inhomogeneity and noise areas by using of the spatial neighborhood information. In this process, the threshold values of up and down cut-off are applied to adjust different membership of pixel. The experiments on the segmentation results demonstrate that the algorithm performs more robust to noise than the standard FCM algorithm and MFCM algorithm.2,In order to overcome the sensitive of FCM algorithm to the initial value and sensitive to noise. This paper proposes two FCM based algorithms with incorporating Monkey-King genetic algorithm and real-coded chaotic quantum-inspired genetic algorithm. Then the fitness function contained neighbor information is set up according to the object function in FCM algorithm. By applying intelligent algorithms, we can achieve the global optimum. The proposed methods can effectively avoid getting into the local optimum solution and more robust to noise. The experiments on the segmentation results demonstrates that the algorithm performs the effective ability of searching global optimal solution and more robust to noise.3,The selection of threshold and noise-immunity play an important role in image segmentation. Then, an improved fast OTSU remote-sensing image segmentation algorithm based on the shuffled frog-leaping algorithm is proposed. We introduce a novel factor H ij incorporating both the local spatial relationship and the local gray-level relationship into the segmentation algorithm. In this method, the improved maximum between-cluster variance is used as the fitness function of shuffled frog-leaping algorithm algorithm. By applying shuffled frog-leaping algorithm to search for the global optimal threshold that realized the classification of target and background. Theoretical analysis and experimental simulation show that the proposed approach greatly enhances the speed of thresholding and has better immunity to noise.