摘要
摄像机标定在光学应用系统中是一个必不可少的步骤,大量的直接信息都来源于图像。为获得精确的摄像头内外参数,提出一种利用改进粒子群算法训练小波神经网络进行摄像机标定的方法。该算法中引入随机粒子群机制,可以有效地克服传统算法收敛速度慢、易陷于局部极小值等缺点。标定实例仿真和分析表明,该算法在收敛速度、计算精度和平均收敛性能方面都有较大改进,可有效确定摄像机的内外参数。
Camera calibration is an essential step in optical applications, a large number of direct information comes from the image. In order to obtain accurate internal and external camera parameters, This paper presents a method using improved particle swarm algorithm to train wavelet neural network for camera calibration. The introduction of random particle swarm algorithm is a mechanism that can effectively overcome in the traditional algorithm, such as slow convergence and so easily caught in local minimum and so on. The simulation and analysis indicates that the algorithm has improved greatly in convergence speed, accuracy and the average convergence performance, which can effectively determine the internal and external camera parameters.
出处
《光学仪器》
2010年第4期6-10,共5页
Optical Instruments
关键词
PSO算法
摄像机标定
随机粒子群
小波神经网络
PSO algorithm
camera calibration
random particle swarm
wavelet neural network
