基于BP算法的双目立体视觉标定技术优化分析  

Optimization analysis of binocular stereo vision calibration technique based on BP algorithm

作　　者：晏芙蓉 李伟 高志远 YAN Furong;LI Wei;GAO Zhiyuan(Institute of Intelligent Manufacturing,Wuhan Optics-valley Vocational College,Wuhan,430077,China;Wuhan Guide Infrared Co.,Ltd.,Wuhan 430077,China;Institute of Intelligent Manufacturing,Guangdong Institute of Technology,Zhaoqing,526100,China)

机构地区：[1]武汉光谷职业学院智能制造学院,湖北武汉430077 [2]武汉高德红外股份有限公司,湖北武汉430077 [3]广东理工学院智能制造学院,广东肇庆526100

出　　处：《中国高新科技》2024年第11期39-41,共3页

基　　金：2022年广东省普通高校青年创新人才类项目(2022KQNCX128)。

摘　　要：摄像机标定是指从多幅二维图像中提取图像的诸多特征,构成三维图像信息。而三维图像重构的逼真程度主要由摄像机标定的精度决定。传统摄像机标定算法中,往往会引入诸多假定的参数,导致计算量比较大,三维重构的精度也比较低。为了从根本上解决上述的缺陷与不足,引入BP神经算法,以此提高摄像机标定的精度和速度,有效增加三维重构的逼真程度,提出基于BP神经网络算法的双目立体视觉标定技术。实验数据表明,通过双目标定技术引入BP神经网络算法后,三维重构的精度有明显提升,算法的收敛速度快,该算法具有较强的正确性和可行性。Camera calibration refers to one of the main ways to extract many features of an image from multiple two-dimensional images,which constitutes three-dimensional image information,and the degree of fidelity of three-dimensional image reconstruction is mainly determined by the accuracy of camera calibration.In the traditional camera calibration algorithm,many assumed parameters are often introduced,resulting in a relatively large amount of computation of the algorithm,and the accuracy of 3D reconstruction is also relatively low.In order to fundamentally solve the above defects and deficiencies,BP neural algorithm is introduced as a way to improve the accuracy and speed of camera calibration and effectively increase the degree of fidelity of three-dimensional reconstruction,and the binocular stereo vision calibration technology based on BP neural network algorithm is proposed.The experimental data show that after the introduction of BP neural network algorithm through binocular calibration technology,the accuracy of 3D reconstruction is significantly improved,and the convergence speed of the algorithm is fast,which verifies that the algorithm has strong correctness and feasibility.

关 键 词：摄像机标定 神经网络 双目立体视觉 优化研究 

分 类 号：O436[机械工程—光学工程]