基于樽海鞘和自适应差分进化的相机内参优化

该文针对相机标定过程中因优化算法所引起的精度不足、稳定性差、易陷入局部最优的问题,提出将樽海鞘优化算法和自适应差分进化算法相结合的相机标定优化算法。该混合算法利用樽海鞘优化算法提高精度,利用自适应差分进化算法增强局部搜索能力,在不同迭代阶段对适应度函数采用分段优化方式,实现平衡局部和全局搜索能力。实验采用每格50 mm×50 mm标准的棋盘格作为标定板,选取15张不同角度的标定图片,图片有效像素为4608 pixe×l3456 pixel,分别利用张正友标定法、樽海鞘算法以及本文提出的樽海鞘-自适应差分进化混合算法进行相机内参的优化。实验结果表明该文提出的混合算法比传统标定方法重投影误差更小,标定精度更高。

Camera Internal Parameter Optimization Based on Salp Swarm Algorithm and Self-adaptive Differential Evolution

Focus on the problems of the camera calibration which make the result of camera calibration not accuracy and stabilized enough,and easy to fall into the local optimum,those caused by the optimization algorithm,we proposed a camera calibration optimization algorithm which combined the salp swarm optimization algorithm and the adaptive differential evolution algorithm together called salp-adaptive differential evolution hybrid algorithm. The hybrid algorithm uses salp swarm optimization algorithm to improve the accuracy,and uses adaptive differential evolution to enhance the local searching ability. In various stages of iteration,the fitness function is optimized in sections to achieve the balance of local and global searching ability. In the process of experiment,the standard chessboard of 50 mm×50 mm is used as the calibration board,and 15 calibration pictures with different angles are selected. The effective pixels of the pictures are 4608 pixel×3456 pixel.We used different algorithm to optimize the camera inner parameters,they are Zhang Zhengyou’s calibration method,salp swarm optimization algorithm and the hybrid algorithm. The experimental results show that the hybrid algorithm proposed in this paper has smaller re projection error and higher calibration accuracy than the traditional calibration method.