光子学报, 2009, 38 (4): 992, 网络出版: 2010-05-10
基于粒子群优化的核匹配追踪目标识别
Object Recognition Using Particle Swarm Optimization based Kernel Matching Pursuit
目标识别 核匹配追踪 粒子群优化 光学信息处理 进化算法 Object recognition Kernel matching pursuits Particle swarm optimization Optical information processing Evolutionary algorithms
摘要
提出了一种基于粒子群优化的用于目标识别的核匹配追踪算法.该算法用粒子群优化算法在基函数字典中选择最优的基函数,大大降低了基匹配追踪算法的计算复杂度.通过与标准核匹配追踪算法及基于遗传算法的核匹配追踪算法对UCI数据集及纹理图像的识别试验表明,核匹配追踪算法优良的分类性能以及粒子群优化算法高效的全局搜索能力使新算法能有效识别目标数据.
Abstract
A method for object recognition using Kernel matching pursuits based on particle swarm optimization technique is presented. By using the particle swarm optimization algorithm in search basic function data in function dictionary, the Kernel matching pursuits method can reduces computational complexity of basic matching pursuit algorithm. As compared with kernel matching pursuits in UCI database and texture image recognition, the proposed algorithm can recognize desired object effectively. This owes to the fact that the kernel matching pursuits is a good classification method and Particle Swarm Optimization algorithm makes use of the global search ability to get the best matching signal structure.
徐小慧, 魏鑫, 张安. 基于粒子群优化的核匹配追踪目标识别[J]. 光子学报, 2009, 38(4): 992. XU Xiao-hui, WEI Xin, ZHANG An. Object Recognition Using Particle Swarm Optimization based Kernel Matching Pursuit[J]. ACTA PHOTONICA SINICA, 2009, 38(4): 992.
PDF全文
