Based On Immune Genetic Algorithm For Image Segmentation Method

Image engineering is a subject that has been developed in recent years,and it has many contents.According to the degree of abstract and the investigate methods, the research on it can be divided into three levels:image processing,image analysis and image comprehension.One of image processing goals is the pattern recognition, but the image segmentation and the survey are the pattern recognition work foundations.The image segmentation is that the image is divided into to the significance some regions,then carried on the description to these regions,equal in withdrawing the characteristic of certain target sector images,at last judged image whether it has the interested goal.Biological immune system is a complicated self-adaptive system,which owns the ability of preventing outside pathogens invading human's body according to lots of mechanism.It owns the abilities like immune memory,antibody's self-recognizing ability and immune diversity. Meanwhile biological immune system shows lots of intelligent characteristics,like recognizing and replying different antigen.Combing with these characteristics, immune algorithm has the abilities of excellent diversity,robustness and invisible parallelism.With the study of biological immune system and development of computer immunology,immune optimization is playing more and more important role in the areas of efficient optimization and intelligent computing.It has been applied in many areas,like pattern reorganization,fault diagnosis and computer safety.However,comparing with other intelligent algorithms,immune algorithm also owns some disadvantage,like premature convergence and poor ability in local optimization.Therefore,the improved research of immune intelligence has become the main and the most welcome area in network,intelligence,control and computer.The immune system is a complex adaptive system,which can effectively use a variety of defense mechanisms for the invasion of external pathogens. Immunological memory for specific performance,antibodies and immune self-recognition of the advantages of diversity.At the same time,the biological immune system in operation to show the many smart features such as on a variety of antigen recognition and response is an evolutionary process of the actual learning process.Therefore,the combination of these characteristics of the artificial immune algorithm,the immune system retains a number of biological characteristics,such as diversity,and robustness,such as implicit parallelism.Biological immune system through a systematic study,at the same time as the computer of the continuous development of immunology,the immune optimization optimization in high-performance computing technology and smart in these two areas,with more and more widely used in pattern recognition,fault diagnosis,in areas such as computer security is widely practical application.However,similar to other new intelligent algorithms,artificial immune algorithm there are some inadequacies,such as the existence of premature convergence,lack of local search capabilities. Therefore,to improve the immune intelligence research,has become the network, intelligent,control,calculation of the areas of focus of the study and one of the hot. This article's main task is to study how the immunization program thinking applies to the issue of image segmentation,to achieve better image segmentation.To this end,this article will focus on how the image segmentation problem is transformed into a constrained optimization problem,and planning the use of immune optimization algorithm to complete the work.This study reads as follows:An overview of the purpose of image segmentation and significance of the study,pointed out that the current method of image segmentation tends to split as a constrained optimization problem.Artificial Immune System,introduced the research background and a variety of immune-based mechanism design principle of the optimization algorithm research.Study of image segmentation are reviewed, introducing the basic concepts of image segmentation and methods,parallel and serial regional segmentation segmentation technology.Introduced the basic theory of genetic algorithms,including the standard genetic algorithm the basic concepts,the basic principles and characteristics.Genetic system to study immune mechanisms and performance,and on the basis of established principles of artificial immune system and its application in optimization problems.Immunization program focuses on the design of algorithm steps.Since the genetic algorithm with global search capability,is an iterative optimization algorithm,and it has some combination of image segmentation methods,both to ensure that the effect of the partition,but also has good real-time, which is widely used.However,due to the complexity of the image itself and a large quantity of issues such as computing,genetic algorithm-based segmentation of the existence of slow and easy to shortcomings such as the formation of premature convergence.And the immunization program as a strong robustness of the random search algorithm,the genetic algorithm based on the increase of the immune operator, able to interfere with the overall strength of a certain parallel search process,a genetic algorithm to overcome the inevitable degradation phenomenon,and in affinity can be maintained under the guidance of the diversity of individuals,to some extent,a genetic algorithm to solve easily into the shortcomings of local optimum, therefore,the immunization program can be seen as a kind of genetic algorithm to improve its application to optimize image segmentation process,the genetic algorithm to achieve more precise than the segmentation results.The threshold method is the most commonly used method of image segmentation,its simple to calculate the volume of small,stable performance in image segmentation as a fundamental and most widely used segmentation.However,appropriate choice of threshold or not the effect of the partition plays a decisive role.In some popular threshold selection algorithm,image segmentation for the control of information loss caused by a criteria based on information entropy image threshold selection method automatically has aroused great concern,the theoretical research and practical applications are was a major breakthrough.Researchers,such as the use of genetic algorithms to efficient optimization algorithm based on information entropy theory of segmentation threshold value,both to ensure that the results,but also has good real-time.However,some special images,such methods can not get the desired effect.For example,the object and the background image for the gray level or gray similar histogram for the multi-peak and trough,such as no major complications, genetic algorithm is used to get the best goal of the threshold can not be clear from the context extraction out due to the genetic algorithm easily trapped into local optimal solution,obtained by maximum entropy threshold segmentation Department may be the object and background gray-scale part of the wrong number of points,the actual optimal threshold appear in the entropy of a local maximum the value of Department.Taking into account the immunization program in the optimization problem of the fine features,focus on the immunization program in the image segmentation in the applied research,discussed image segmentation of a few basic steps,including pretreatment,regional segmentation and connectivity markers,post-processing (morphological filter).Based on the theory of maximum entropy image threshold segmentation algorithm as follows:invasion of antigens and encoding,initialization, extraction of vaccines,stands to determine the calculation of fitness,antibody groups genetic operator to operate(based on individual concentration of inhibition and promotion of adaptive selection operator,crossover operator,mutation operator),the immune operator to operate,update groups.The use of quantitative image histogram to calculate the entropy,can greatly reduce the computational complexity.To 256 gray-scale image as an example,after quantification of 32 gray-scale,therefore, based on the image histogram to quantify the amount of computing only the original image histogram-based computation of 1 / 8.Immunization program for the largest information-based image segmentation algorithm theory applications in simulation and with the corresponding comparison of the classic genetic algorithm,the segmentation results and experimental data has been analyzed and summarized.The experimental results show that image-based immunization program threshold segmentation,not only can automatically find a suitable threshold,and the adaptive algorithm is not only suitable for the structure with a bimodal histogram of the image, the more complex structure of the histogram of the image also very effective.In this paper,the immune programming algorithm,using local feature information structure immune operator,through vaccination and immune selection, to a certain degree of intensity of intervention in the overall search process in parallel,effectively curbed in the past,the degradation of genetic algorithms is to construct algorithms rapid global convergence with good performance,validate solutions to complex problems in the use of local feature information to find the global optimal solution of the feasibility and effectiveness.Affinity and,under the guidance of the immunization program to maintain a certain degree of individual diversity,to avoid issues such as premature convergence and improve the overall performance of the algorithm.The need for more complex problems,the superiority of the immunization program the greater,and the image segmentation inherent complexity and computing features such as volume,so that the advantages of the immunization program has been brought into full play.