Research Of Biogeography-based Optimization Algorithm And Its Application In The Image Segmentation

In recent years, Along with many kinds of bionic and intelligent optimization algorithmput forward and applied widely, this new kind of optimization algorithm which follows theexample of nature’s evolutionary rules causes many researchers’ attention. They become animportant research direction of the optimization algorithms and their respective advantages inpractical application are more and more outstanding. So more and more people are sure oftheir value of research. At present, the intelligent bionic algorithms have achieved successfulapplications in optimization, pattern recognition, automatic control, signal and imageprocessing, biomedical engineering, electrical engineering, mechanical engineering,communication engineering, economic management and other areas.Image segmentation is the key step in image engineering. Its principle is extracting theobjects or interesting regions from the background and its effect affects the image quality. Soimage segmentation is a hot field which is in the development and there is not an universalsegmentation algorithm and evaluation criteria now. Image segmentation and extraction arewidely used in many fields, such as medical, military, industrial automation, process control,document image processing, security monitoring, online product inspection, agriculturalengineering, remote sensing and other fields inseparable from the image segmentationtechnology.Fuzzy C means (FCM) algorithm is one of the methods of clustering analysis and it isthe most widely used unsupervised classification algorithm currently. However, this algorithmhas obvious disadvantages such as setting the number of clusters in advance, being verysensitive on the choice of initial cluster centers, easily falling into local minimum value andso on.In this paper, biogeography-based optimization algorithm (BBO) is introduced in detailand it is in the growth stage which is relate to the application field less. So the innovation ofthis paper is combining BBO algorithm and fuzzy C means algorithm to form a new mixedclustering algorithm (BBO-FCM) which will be used in image segmentation field. Theprinciple of the new algorithm is using the advantage of BBO algorithm to fill thedisadvantage of FCM algorithm. BBO algorithm has strong global search ability and use anumber of habitat migration to optimize the FCM clustering criterion function, so this waycan overcome the problems that FCM algorithm is sensitive to initial value and easy to fallinto local optimum. The mixed algorithm also retains the mutation and elitist operations of BBO algorithm which enhance the searching capability; in addition, the traditional FCMalgorithm for image segmentation can not make full use of the information betweenneighborhood pixels, and the BBO algorithm migration operation can solve this problem.This paper not only designs the BBO-FCM algorithm, but also designs the PSO-FCMalgorithm, AFSA-FCM algorithm and ABC-FCM algorithm. Then the four algorithms areapplied to iris data sets and wine data sets for a preliminary inspection of these four kinds ofclustering algorithm.The experiment proves that the BBO-FCM algorithm has obviousadvantages in this respect. The last chapter uses the above four algorithms and the originalFCM algorithm on eight different groups of segmentation, from the qualitative andquantitative dual analysis results, the new mixed optimal algorithm (BBO-FCM) shows astrong advantage and the effect is better than traditional FCM algorithm and several otherbionic optimization algorithms used to contrast.