Research On Application Of Artificial Fish-Swarm Algorithm In Image Processing

The automation and intelligence of image processing and analysis is a hot and key problem in the filed of image processing and analysis. Artificial Fish-Swarm Algorithm (AFSA) is a group of animal behavior-based intelligent optimization algorithm, which is a typical behavioral application of artificial intelligence. In addition, it has intelligence properties such as high parallel adaptation, self-organizing and synergy. As well, it can effectively overcome the local optimal value, has the ability to obtain the global optimum. Therefore, this thesis employs AFSA to handle some problems in image processing and analysis and attempts to explore an effective way to apply the algorithm to image processing.First of all, this paper gives an introduction of the principle, model, application and modification of AFSA. Then, this paper places emphasis up on application of the algorithm in image enhancement and segmentation, which are two significant points in image processing. In general conventional enhancement algorithms need much manual intervention, which lack of adaptability, meanwhile, some image enhancement algorithms are adaptive, but these methods need a great amount of calculation. In order to improve the calculation efficiency, this paper makes use of AFSA to seek the optimal parameters for image enhancement automatically, it can greatly improve the calculation efficiency, obtain good image enhancement effect.Image segmentation is the key step of image processing and analysis and thresholding segmentation is a widely used method in regional segmentation. In practical applications, threshold value often is selected by computer automatically. It is often difficult to obtain a satisfactory segmentation result by using thresholding method based on one-dimensional histogram threshold for it is does not use of space information. The segmentation method based on two-dimensional histograms can achieve better segmentation results; however conventional two-dimensional thresholding methods are time-consuming. In fact, most of automatic threshold selection methods are to search the best gray value which can make the objective get extreme value and takes that gray value as the optimal thresholds; in essence, it is an optimization problem. Artificial fish algorithm can solve this kind of problem; hence this thesis proposes to take advantage of AFSA to hunt the optimum solution, hoping that AFSA can get the optimum two-dimensional thresholds quickly.In summary, the main work of this paper is that it has applied AFSA to some image processing problems such as the enhancement, segmentation, corresponding experimental results proves that AFSA has good feasibility and effectiveness in the field of image processing and analysis, which has a good application prospect.