基于立体标定件的双目视觉目标测量方法

2020-08-13 07:17:51蓝畋华云松
软件 2020年6期

蓝畋 华云松

摘  要: 为了实现双目摄像机对目标物测量精度的提升,本文提出一种基于立体标定件的双目立体视觉目标物测量方法。在摄像机标定步骤中,本方法在传统标定方法基础上,利用立體标定件的几何特性以及相邻面的图像冗余对初步标定的内外参矩阵进行修正,提升标定数据的精确性,然后利用SGBM(Semi-Global Block Matching)立体匹配算法对校正后的左右图像进行匹配,获取测量物上各点的三维坐标值,从而实现对目标物的尺寸测量。分别用不同的物体进行测量后,实验结果证明本方法能将误差控制在1 mm以内,具有较好的准确性。

关键词: 双目立体视觉测量;立体标定件;相机标定;立体匹配

中图分类号: TP317.4    文献标识码: A    DOI:10.3969/j.issn.1003-6970.2020.06.020

本文著录格式:蓝畋,华云松. 基于立体标定件的双目视觉目标测量方法[J]. 软件,2020,41(06):9195

【Abstract】: In order to improve the measurement accuracy of the target by the binocular camera, a binocular stereo vision target measurement method based on stereo calibration parts is proposed in this paper. In the camera calibration step, based on the traditional calibration method, this method uses the geometric characteristics of the three-dimensional calibration part and the image redundancy of adjacent surfaces to modify the initial calibration internal and external parameter matrix to improve the accuracy of the calibration data. The Semi-Global Block Matching algorithm matches the corrected left and right images to obtain the three-dimensional coordinate values of the points on the measurement object, thereby achieving the measurement of the size of the target object. After measuring with different objects, the experimental results prove that the method can control the error within 1mm, and has better accuracy.

【Key words】: Binocular stereo vision measurement; Stereo calibration; Camera calibration; Stereo matching

0  引言

人眼能够轻松的感知物体在空间中的位置,但无法实现对三维空间物体的精准测量。立体视觉作为计算机视觉的重要分支,它利用图像采集及预处理、摄像机标定、立体校正、立体匹配、图像深度信息还原等步骤实现对目标物的非接触精准测量。相比于人工测量,双目立体视觉测量不仅有效率高、精度高、自动化程度高等优点,更能获取目标测量物的特征几何参数及空间姿态特性,实现了目标物测量的智能化。

本文在传统张正友标定法[2]的基础上,根据立体标定件的几何特性及标定件相邻面的图像冗余作为约束,对双目摄像机内外参进行修正,再结合立体校正和SGBM立体匹配算法实现对目标物的高精度测量。……

登录APP查看全文