Study Of Image Segmentation Based On Clone Selection And Particle Swarm Algorithm

Image segmentation is a key technology in digital image processing which is used for the next step of image processing. Accordingly its performance have a direct impact to the visual effects. Image segmentation is a hot technology which let many scholars pay attention to it, and many methods of this technology have developed rapidly. Accordingly it had been widely applied in many fields, such as transportation field, bio-medical field, military field, industrial control field etc.Threshold segmentation is one of the most commonly used method of image segmentation. The object and background classification is the pith of threshold segmentation. The selected threshold determined the quality of image segmentation, so how to select the appropriate threshold became the key technology of threshold segmentation. There are many methods to select the threshold. The algorithm of threshold segmentation based on entropy is a better method of image segmentation which took into account the effective image information.This paper analyzed the one-dimensional and two-dimensional maximum entropy image segmentation. The one-dimensional maximum entropy image segmentation method is just used the image pixel gray distribution information while ignoring the spatial information of image pixels. This method have a weak anti-noise capacity and a poor robustness. The two-dimensional maximum entropy image segmentation method not only pay attention to the distribution information of image pixel gray but also took into account the spatial information of pixels. This method can improve the noise immunity.The solving process of threshold segmentation based on image maximum entropy was actually the process of finding the optimal solution for segmentation function. The clonal particle swarm optimization algorithm is considered as a new global optimization search algorithm which have fine memory function and better convergence. Clonal particle swarm optimization algorithm can be used to maximum entropy image segmentation which finding the optimal solution of the segmentation function to obtain the best threshold. Clonal particle swarm optimization algorithm is proposed based on the characteristics of clonal selection and particle swarm algorithm. It is combined the advantages of two algorithms and have overcome their shortcomings so that optimized the performance of the algorithm. This paper is used clonal particle swarm optimization algorithm for one-dimensional and two-dimensional maximum entropy image segmentation. Accroding to lots of experiment, the simulation results show that the algorithm can accelerate convergence rates of image segmentation and obtain optimal threshold for image segmentation quickly and can obtain the better results of image segmentation.