摘要
主动轮廓线模型凹陷边界点的寻优属于动态优化问题,由于其复杂性,传统方法不能准确搜索到最佳边界点。若采用单一的粒子群优化算法求解,不仅耗时,而且容易陷入局部极值。针对以上问题,提出一种多种群粒子群优化算法,并将其应用于主动轮廓线模型的边界寻优过程中。该算法为每个控制点设置一个种群,各种群之间通过共享信息的方式协作寻优,从而避免采用单一PSO算法容易早熟的缺点,同时扩大了控制点的搜索区域,提高了收敛速度。将改进方法与传统方法进行了对比,实验结果证明了改进方法的有效性。
Searching for the object＇s concavities of active contour model （Snake model） is a dynamic optimization problem. Due to its complexity, the traditional active contour model and single particle swarm optimization algorithm converge slowly and easily converge to local optima. Aiming at the above problems, multi-swarm PSO algorithm was proposed to apply in active contour models for the sake of expanding the control point of the searching area and optimizing convergence speed. In this algorithm, every control point was corresponding to one swarm. Sub-swarms worked together via information sharing. The results obtained by the multi-swarm PSO algorithm have been compared with that of single PSO algorithm and the conventional method. Experimental results show that the proposed algorithm is more efficient.
出处
《计算机应用》
CSCD
北大核心
2008年第10期2622-2624,2627,共4页
journal of Computer Applications
基金
甘肃省自然科学研究基金项目(0803RJZA025)
关键词
多种群
粒子群优化算法
蛇模型
主动轮廓线模型
图像分割
multi-swarm
Particle Swarm Optimization （PSO）
Snake model
active contour model
image segmentation
