基于蚁群算法的二维最大熵分割算法.pdfVIP

第 36卷第12期 光 子 学 报 Vo1．36 No．12 2007年 12月 ACTA PHoToNICA SINICA December 2007 Two-Dimensional Maximum Entropy Segmentation Based on Ant Colony Optimization CAO Zhan—hui，LI Yan—j un，ZHANG Ke (College of Astronautic，Northwestern Polytechnical University，Xi an 710072，China) Abstract：The 2-D maximum entropy method reflects information of the gray distribution and space—related information of the neighborhood． Therefore the segmentation result is more accurate than the 1-D method．However its computational cost is an obstacle in application．Ant Colony Optimization is has been successfully applied to some discrete problems，such as the traveling salesman problem．The ant colony optimization is introduced and the 2-D maximum entropy segmentation is presented based on ant colony optimization．Through the experiments of segmenting infrared images，it is about 60 times faster than the traditional exhaustive search algorithm．The proposed algorithm has been proved to be fast，simple and effective． Key words：Image segmentation；Ant colony optimization；Two—dimensional maximum entropy； Thresh0】d CLCN：TP391．41 Document Code：A Article ID：1004—4213(2007)12-2377—4 0 Introduction average information． The maximum entropy method is successfully applied in the image Image segmentation is a very important reconstruction．It is also used to select the optimal process in image processing and has been the threshold to segment images．2-D MES is based on subject of intensive research． The threshold the 2-D histogram of the