基于遗传算法的摄像机自标定方法  

Self-calibration for Cameras Based on Genetic Algorithm

作　　者：郭秋艳[1] 刘鹏飞[1] 安平[1] 张兆杨[1] 

机构地区：[1]上海大学通信与信息工程学院,上海200072

出　　处：《中国图象图形学报》2006年第11期1712-1715,共4页Journal of Image and Graphics

基　　金：上海市自然科学基金项目(04ZR14056);上海市教委发展基金项目(04AB56);国家自然科学基金项目(60572127)

摘　　要：摄像机标定是计算机视觉领域的关键技术,其中的自标定是只根据图像计算摄像机的内参数,其标定过程简单,适用性强。由于传统的用于摄像机自标定的Kruppa方程不仅需要计算基础矩阵,还要计算图像的极点,而图像的极点又不是固定不变的,且会导致计算结果的不稳定,为此,针对传统摄像机自标定方法的上述不足,利用遗传算法完成了Hartley新的Kruppa方程的摄像机自标定过程,以便将这个过程完全转化为通过代价函数最小化来求得摄像机的内参数,这就排除了极点的不稳定因素。实验结果表明,该方法是简单、有效的,可以作为一种通用的标定工具。Camera calibration is a key technology in computer vision, in which camera self-calibration technology is to compute the camera＇ s intrinsic parameters from a series of images. Compared to traditional camera calibration methods, the process of self-calibration is simpler and more convenient for application, The self-calibration technology using Kruppa equation not only requires computing the fundamental matrix, but also computing the epipoles of images which are variable with the different images and will result in unstable computation results. GA algorithm is used to complete the self- calibration processing by estimating the new and simple Kruppa equations defined by Hartley. At last, the self-calibration problem is converted into the minimization of the cost function, so that the epipoles instability is eliminated and the calibration effect is improved, Experimental results show that the proposed method is simple and effective, and can become an versatile tool for caiuera calibration.

关 键 词：摄像机自标定 KRUPPA方程 基础矩阵 遗传算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]