Image Segmentation Method Based On Improved Genetic Algorithm

In order to automatically determine the optimal threshold in image segmentation，this paper presented a new method of image segmentation based on improved geneticalgorithm，it used this improved genetic algorithm to globally optimize2-mension Otsuimage segmentation functions．Now the field of image processing There are many typesof segmentation algorithms in general can be divided from the types of basicsegmentation based on region segmentation and edge detection method based on thesegmentation method. In practice, the application form the main types are:threshold-type, edge detection-and track-type region.With technological development, the following of some formation show newtheory. Since this different focus, so the main type introduced threshold segmentationmethod. Threshold method is experimentally proven to be very effective method imagesegmentation threshold extraction method The principle is the destination portion ofimage to be processed portion of the gradation characteristics of the impurity on thecomparison, understanding image for a number of varying gray feature separate part ofthe value that is determined segmentation threshold, the purpose of image blocks andimpurities image block separated. Convert image threshold segmentation method thenumber of channels, convert grayscale image to binary image, reducing the amount ofstorage, reducing the threshold to strike and the calculated amount of computation.In order to strike a better threshold image segmentation, we propose a put based onimproved genetic algorithm and improved image segmentation Ostu Combination of thenew method of image segmentation. Genetic algorithm is a reference to biologicalreproduction of human development theory Evolution born field of image processingtheory. In the process of biological evolution, individual organisms adapt to theenvironment is through heredity, variation, and gradually phase out low-qualityindividuals to retain excellent individuals. GA is a reference to the above phenomenon.By simple traditional genetic algorithm, Otsu image segmentation method and the likeconventional image segmentation target image and the presence of the entire image arearatio uncertainty problem, the solution of the problem for a long time to strike easily formed in advance dummy solution, resulting in image segmentation threshold strike isnot simple. Therefore, this problem is in the basic research goal based on the geneticalgorithm, the threshold strike solve the problem for a long time, is difficult to form thepseudo-solution. And image segmentation should be clear, accurate, clear edges, noextra background image. Objective is to use genetic population and degree of themonomers of all monomers modified discrete exponential algorithm to set the parametervalues to assure that all the monomers to maintain its complexity and plumpness, thethreshold value can reduce the time to strike, it is difficult to form pseudo-solution, theoptimal solution to form a stable, so as to achieve a satisfactory image segmentationresults.This method could automatically adjust the parameters of genetic algorithmaccording to the fitness values of individuals and the decentralizing degree ofindividuals of the population，and kept the variety of population for rapidly convergingto get the optimal thresholds in image segmentation，it overcame the shortcomingsincluding worse convergent speed，easy to premature that exist in traditional geneticalgorithm etc．The theoretically analysis and simulate experiments show that the rangeof the thresholds is more stable and it consumes less time greatly and better satisfies therequest of real-time processing in image segmentation by using this new method，compared with2-mension Otsu image segmentation and genetic algorithm based imagesegmentation．...