眼科手术机器人双目视觉标定方法研究
2019-12-16
作者：闫兴1，曹禹1，王晓楠2，朱立夫2，王君1，何文浩2
单位：1北京信息科技大学；2中国科学院自动化研究所
摘要：双目视觉系统通过相机模拟人眼，基于视差原理，从不同角度采集被测物体的两幅图像获取其二维信息，建立特征点的对应关系，计算出位置偏差，从而实现三维重建。眼科手术机器人双目视觉系统选用的标定方法是以张正友摄像机标定方法为基础，利用Harris角点检测方法实现亚像素点的精确化，对采集图像进行畸变矫正。将得出的标定结果与常用传统标定方法MATLAB相机标定工具箱Toolbox_calib的标定结果进行比较，分析像素误差。结果表明：使用新算法标定后的结果更加稳定，像素误差更小，精度更高，适用情况更加广泛，可用于眼科手术机器人的双目视觉标定系统中。
关键词：眼科手术；双目视觉；相机标定；角点检测；opencv
中图分类号：TG713；TH741文献标志码：ADOI：10.3969/j.issn.1000-7008.2019.12.023
Research on Binocular Vision Calibration Method for Ophthalmic Surgery Robot
Yan Xing，Cao Yu，Wang Xiaonan，Zhu Lifu，Wang Jun，He Wenhao
Abstract:Binocular vision system simulates the human eye through two cameras.Based on the parallax principle，two images of the measured object are collected from different angles to obtain its twodimensional information.By establishing the corresponding relationship between the feature points，the position deviation is calculated，and the threedimensional reconstruction is realized.The calibration method for the binocular vision system with ophthalmic surgery robots is based on Zhang Zhengyou′s camera calibration method，the Harris corner detection method is used to realize the accuracy of subpixel points，and correcting the distortion of the collected image，the calibration results obtained are compared with those obtained by traditional calibration methods such as Toolbox_calib，a calibration toolbox for camera calibration in MATLAB.Analyzing the error of the pixel，the result after calibration with the new algorithm is more stable，the pixel error is smaller，the precision is higher，and the application is more extensive，and this method can be used in binocular vision calibration system of ophthalmic surgery robot.
Keywords:ophthalmic surgery；binocular vision；camera calibration；corner detection；opencv