Research On Image Segmentation Algorithm Based On Fuzzy Clustering

Image segmentation is the first step in the understanding and analysis of images.It is closely related to our life and has a more and more profound impact on our life.With the continuous development of computer technology,image segmentation also presents a new look.Therefore,it is of profound significance to study the image segmentation method and make it better serve people's life.There are various methods of image segmentation,each method has its own suitable scene or image type,it is necessary to select a specific method for a specific image.Fuzzy C-means(FCM)algorithm has attracted extensive attention due to its characteristics of simplicity,easy to understand and strong local search ability.However,FCM algorithm also has some defects or deficiencies.The traditional FCM algorithm is based on the grayscale of image pixels,which is unreasonable to some extent.The accuracy of image segmentation,the quality of segmentation and people's visual effects are often very different from people's expectations.Because this method is sensitive to noise and is easily affected by the initial value,the algorithm converges to the local extreme value,resulting in the performance of the algorithm is reduced and the effect of image segmentation is worse.In view of some problems existing in FCM image segmentation method,this thesis combines some other methods to improve FCM,and applies them in image segmentation to verify the performance of the algorithm,so as to make the performance of the algorithm better and the effect of ima ge segmentation better.The main work of this thesis is as follows:1.Improve the FCM image segmentation algorithm by combining markov model.Since the markov model can well represent the spatial position relationship between pixels,it has good noise resistance.By changing the objective function of FCM algorithm,the prior probability of markov model is introduced into the objective function as the correction,thus to change the purpose of the objective function and membership degree matrix calculation,and constraint coefficient to regulate constraint ability,in order to enhance the noise resistance and segmentation precision of the algorithm,and through the corresponding experiment to test the performance of the algorithm.2.Aiming at the problems of slow convergence speed,low accuracy and low efficiency of genetic algorithm,the hybrid optimization design and implementation are carried out.The method of real number coding and the genetic operator of other genetic algorithms are adopted to optimize the design and implementation,that is,the selection based on fitness ranking and the adaptive strategy of solving crossover and mutation probability make the actual effect of the algorithm is better.3.Genetic algorithm is used to determine the initial clustering center of FCM algorithm,and an improved FCM image segmentation method based on genetic algorithm is proposed.The combination of genetic algorithm and fuzzy clustering improves the performance of FCM and significantly reduces the number of iterations.Using genetic algorithm to determine the initial clustering center of fuzzy clu stering,the improved FCM algorithm was used for image segmentation.Combining the advantages of the two algorithms,the image segmentation performance of fuzzy clustering algorithm is improved on the whole.