Research On Image Segmentation Algorithm Based On Improved Fuzzy Clustering

Image information is an important form of people's perception of the world.With the development of artificial intelligence,people have higher and higher requirements for image processing.In the entire process of image processing,image segmentation is in a relatively basic position.Accuracy has a great impact,so image segmentation plays an important role in engineering and daily life.For a long time,researchers have conducted in-depth research in the field of image segmentation,and the methods of image segmentation have become more and more diversified.However,there is still no universal method,so the exploration of image segmentation has been continuing.Among them,the fuzzy C-means(FCM)algorithm has become the method chosen by many scholars for image segmentation because of its simple principle and close to reality.However,the traditional FCM algorithm also has many shortcomings.For example,the clustering center of the algorithm is randomly initialized,which leads to poor global optimization of the algorithm,and the algorithm is more sensitive to noise in the image.In response to the above problems,after analyzing the principle of the FCM algorithm,this article puts forward the following innovations:(1)In order to avoid the algorithm falling into the local optimal value,this paper proposes an adaptive artificial fish school algorithm with a hybrid teaching-learning algorithm,which uses the strong global optimization ability of the artificial fish school algorithm to cluster the FCM algorithm.The center performs optimization to optimize FCM image segmentation.The idea of the algorithm is as follows: First,a fixed field of view and step length are used for the standard artificial fish swarm algorithm.This mode may cause the algorithm to oscillate near the extreme point in the later stage of optimization,thereby missing the optimal solution and reducing the optimization speed.Therefore,this article adjusts the field of view and the size of the step length by introducing the visual step coefficient in the parameter adjustment.As the algorithm progresses,the field of view and step length of the artificial fish become a variable value.Secondly,in terms of mixing with other intelligent algorithms,this article has also improved another intelligent algorithm by teaching-learning algorithm,and its teaching factor It becomes an adaptive value,and then combines the communication process in the teaching-learning algorithm with the idea of differential evolution,combining the adaptive field of view and the step-length artificial fish school algorithm to improve the teaching-learning algorithm,and finally obtains a An improved artificial fish school algorithm with higher optimization speed and accuracy.The objective function of the FCM algorithm is used as the fitness function.The image segmentation comparison experiment is carried out.From the experimental results,compared with the standard FCM algorithm and other improved FCM algorithms,the iteration time of the algorithm is reduced by about 14%-55% on average,and the segmentation accuracy is increased by about 10%.(2)In order to solve the shortcoming that the standard FCM algorithm is very sensitive to noise,this paper proposes an FCM algorithm based on the Markov random field model of the improved potential function.First,the reason why FCM is sensitive to noise is analyzed,and the Markov random field model is introduced.The potential function determination in the standard MRF model is changed into an adaptive method,which strengthens the intensity relationship between pixels in the image.MRF considers the spatial neighborhood information between pixels,can better identify and segment the noise points and discontinuities in the image,and effectively suppress the noise points in the algorithm.Finally,the improved algorithm is used to carry out the anti-noise detection test,and compared with other improved anti-noise FCM algorithms,it proves that the algorithm has better and stable anti-noise performance.Finally,this paper combines the two algorithms proposed above,complements the advantages of the two algorithms,and proposes an improved artificial fish school algorithm based on Markov random field to optimize the FCM image segmentation method,and adds noise to the image Through subjective evaluation and objective data analysis,it shows that the new algorithm not only has high segmentation accuracy,but also has good anti-noise performance without slowing down the segmentation speed.