Research On Image Processing Based On Bionic Intelligence Optimization Algorithms

With the rapid development of the image processing technology, it has been more and morewidely used in the fields of military, medicine, industrial production, telemetry and telecontrol,etc. The complexity and diversity of image information characteristic has become more and moreobvious, and the processing to image information has become more and more difficult too. Onthis accounts such as uncertainty of image information and difficulty in modeling, the raditionaloptimization methods in solving image processing problem become helpless.Bionic intelligentoptimization algorithm is a kind of target optimization algorithm simulating biological behaviorand it is an efficient method for solving complex optimization problem. There is finedevelopment prospect for using bionic intelligence optimization algorithm in image processingproblem. In this paper, the theory and methods that the optimization algorithm of artificial beecolony, bacteria and seeker are used to image processing are studied. Some new methods andway of thinking of image processing are proposed. The main work can be expressed as follows:1. A vector quantization image compression algorithm based on artificial bee colonyoptimization was proposed. Mean squared error was used as objective function which couldmeasure the quantization of codebook, and artificial bee colony optimization algorithm was usedto optimization design of the codebook in vector quantization image compression. In order tofurther improving the performance of the aigorithm, a colony initialization method based onchaotic mapping and reverse learning was used to generate the initial code book, which reducedthe influence of the intial codebook to the optimization results; The mutation operation indifference evolution was use to the search strategy of the artificial bee colony optimization,which speed up the convergence speed of the algorithm; An ideal about a fast codebook searchwhich based on sum of vector was used in the process of calculating fitness function, whichreduce computational effort of the algorithm. Simulation results show that the algorithm hastime-saving calculation and rapid convergence, and the codebook generated by this algorithm is of high quality and good universal.2. A blind image separation algorithm based on bacterial chemotaxis was proposed. Theabsolute value of normalized fourth-order cumulant was used as objective function for blindseparation and the bacterial chemotaxis algorithm was used to optimize it. The separation modelof sequential extraction was used in the separation process and the separated source imagecomponent was wiped off from the mixed images after every extraction process. All the originalimages can be separated out finally. Simulation results show that the algorithm proposed canachieve the blind separation for multiple mixed natural images efficiently. The separationproperty is excellent according to the PSNR value and correlation coefficient value between theseparated image and original image.3. A moving target detection algorithm based on seeker optimisation algorithm wasproposed. The moving target detection problem was transformed into the blind image separationproblem and it can be solved using independent component analysis method. Negentropy wasused as the objective function for blind separation. The excellent optimization ability of seekeroptimisation algorithm was used to optimize the objective function and the separation image ofone independent component can be got. The multiple images can be recovered successfully byrepeating the separation and deflation process. Finally, the clear trajectory of the moving targetcan be obtained successfully.