【摘要】 针对单目直接法视觉里程计在光照变化大的场景下定位精度低、定位鲁棒性差的问题,文中提出一种融合光度参数估计的单目直接法视觉里程计。首先通过对图像帧进行特征匹配实现特征跟踪,建立起图像像素点之间的关联,接着利用非线性优化方法进行光度参数估计,然后在光度参数估计提供的光度参数先验信息的基础上对图像灰度值进行校正,最后利用校正后的图像采用单目直接法视觉里程计对相机运动进行估计。在EuRoC数据集上的实验结果表明,与传统方法相比,文中提出的算法提高单目直接法视觉里程计对光照变化的鲁棒性,定位精度平均提高约18%。更多还原

【Abstract】 Aiming at the problems of low positioning accuracy and poor positioning robustness of the monocular direct method visual odometry in the scene with large illumination changes, a monocular direct method visual odometry integrating photometric parameter estimation is proposed. Firstly, feature tracking is realized by feature matching of image frame, and the correlation between image pixels is established. Secondly, the photometric parameters are estimated by nonlinear optimization method, and then the gray value of image is corrected based on the prior information of photometric parameters provided by photometric parameters estimation. Finally, the monocular direct method visual odometry is used to estimate the camera motion based on the corrected image. Experiments on the EuRoC datasets show that compared with the traditional method, the proposed method improves the robustness of the monocular direct method visual odometry to the change of illumination, and improves the positioning accuracy about 18%.更多还原