Research On Superpixel Image Segmentation Algorithm Based On Fuzzy Clustering

Image segmentation refers to the classification of the original image into different areas with different characteristics of its pixels,and then the target area we need is extracted from these different areas.This is a very important branch in the field of image processing,and it is widely used in all aspects of life.Since fuzzy clustering analysis was introduced into the field of image processing,it has a good solution to the uncertain factors in image segmentation,and the effect is better than other traditional segmentation algorithms.Against disadvantages of classical fuzzy clustering algorithm,based on the image segmentation algorithm was improved,based on the LAB for space weighted gradient watershed image segmentation algorithm,and strengthen the combination of fuzzy clustering based on again strengthen the LAB space under the fuzzy clustering of pixels segmentation algorithm,and through experiment this paper put forward a new algorithm of image segmentation effect is good.The main research work of this paper is as follows:(1)The second chapter mainly introduces the basic ideas and related formulas of Fuzzy C-Means Clustering.In this chapter,the traditional fuzzy clustering algorithm FCM,FCM＿S algorithm,FCM＿S1 algorithm,FCM＿S2 algorithm,En FCM algorithm and FLICM algorithm are introduced respectively.The third chapter introduces the basic theory of watershed segmentation algorithm and Canny operator,and shows the advantages of Canny operator intersecting Sobel through experiments.(2)In Chapter 4,a weighted watershed segmentation algorithm combined with morphology is proposed.By improving the solution of gradient image,firstly,the image in RGB space is converted to LAB space by using the advantage of LAB space.The second step,through the study of traditional Canny operator,consider adding two diagonal directions;Thirdly,the improved Canny operator is used to solve the gradient image of the three channel images in LAB space.Finally,the final gradient image is obtained by information entropy weighting.In the fourth step,the gradient image is reconstructed by morphological gradient,and finally the segmentation result is obtained by watershed segmentation algorithm.According to the subjective observation of the experimental results and the analysis of the objective data,it can be proved that the algorithm proposed in this paper can effectively solve the problem of image segmentation caused by the complex environment of the image,the texture of the object,light shadow or reflection and other factors,such as over-segmentation and under-segmentation.(3)In Chapter 5,En FCM segmentation algorithm combining weighted watershed is proposed,which combines enhanced fuzzy clustering segmentation algorithm with watershed segmentation algorithm based on weighted gradient image in LAB space.By comparing the experimental results,the algorithm in this paper can avoid over-segmentation and also take into account spatial information and gray information of the image.The results of subjective evaluation and objective evaluation showed better results.