摘要
点匹配问题一直是计算机视觉、模式识别、医学临床诊断领域的一项重要的基础性工作。本文提出了一种基于粒子群优化算法的准确、快速和鲁棒性的点匹配方法。该方法首先确定两个特征点集的点匹配问题的能量函数,通过最小化该能量函数可以同时得到点集之间的匹配矩阵和映射参数,利用粒子群优化算法求解变换参数,实验表明,该算法适用于点匹配,具有操作方便、可靠性好、不易陷入局部极值等优点。
The matching of two point-sets plays an important role in computer vision, pattern recognition and medicine diagnose. In this paper, we propose an accurate and robust algorithm for solving the point matching problem using particle swarm optimization. At first, an energy function describing the problem is defined. Secondly, PSO is used to minimize the above energy function, and then we are able to combine the estimation of both spatial mapping parameters and matching matrix between the two point-sets. The experimental results demonstrate the algorithm is simple and reliable, and avoids local extrema.
出处
《系统仿真学报》
CAS
CSCD
2004年第8期1686-1688,1691,共4页
Journal of System Simulation
基金
国家自然科学基金(50275019)
教育部博士学科点专项科研基金(20010441005)。