基于改进遗传算法的基础矩阵估计方法  

Method for Estimating Fundamental Matrix Based on Improved Genetic Algorithm

作　　者：李宏言[1] 盛利元[1] 闻姜[1] 娄继冰[1] 

机构地区：[1]中南大学物理科学与技术学院,长沙410083

出　　处：《计算机工程与应用》2006年第34期57-59,91,共4页Computer Engineering and Applications

摘　　要：基础矩阵估计是计算机视觉的基础。提出了一种新的基于改进遗传算法的基础矩阵估计方法,它以8个匹配点作为最小匹配点集组成染色体,并利用改进8点法估计出相应的基础矩阵,同时以对极距离为核心构造适应度函数,并对传统遗传算法进行了优化改进。实验证明,此方法在估计精度和鲁棒性等方面有很大提高,易于实际应用。Fundamental Matrix Estimation is the key content of Computer Vision.A new method for estimating Fundamental Matrix based on Improved Genetic Algorithm is proposed.It uses eight points to form each chromosome as the minimum subset of correspondences ,and then gets the corresponding Fundamental Matrix by improved 8-point algorithm. Besides,the illness function is constructed based on the distance between correspondences and epipolar lines,and some genetic operators have been improved and optimized.The experiment results show that it is more precise and robust,and easier to implement.

关 键 词：基础矩阵 对极几何 改进8点法 改进遗传算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]