仓库三维重建系统中改进双目匹配SAD算法研究  

IMPROVED SAD ALGORITHM OF BINOCULAR STEREO MATCHING IN WAREHOUSE 3D RECONSTRUCTION SYSTEM

作　　者：刘学君[1] 常梦洁 孔祥旻 晏涌 李柏灿 龚泓博 Liu Xuejun;Chang Mengjie;Kong Xiangmin;Yan Yong;Li Baican;Gong Hongbo(Beijing Institute of Petrochemical Technology,College of Information Engineering,Beijing 102617,China)

机构地区：[1]北京石油化工学院信息工程学院,北京102617

出　　处：《计算机应用与软件》2023年第7期180-184,214,共6页Computer Applications and Software

基　　金：工业和信息化部工业强基专项子课题;北京市级大学生研究训练计划基金项目(2020J00182,2020J00183,2020J00185);2018年第二批产学合作协同育人项目(201802040020)。

摘　　要：利用双目立体视觉技术实现危险化学品仓库的三维重建是库内安全监测的有效手段,立体匹配是其中的重点和难点。提出一种利用加权SAD算法和灰度共生矩阵的双目立体像对稠密匹配算法,该算法对传统SAD匹配算法做出改进,引进服从二维高斯分布的加权系数求出初始视差图,然后结合灰度共生矩阵的4个纹理特征,利用图像纹理信息对空洞进行填充,得到了优化的视差结果。实验结果表明,该算法在标准数据集及仓库模型具有较好的适用性。Using binocular stereo vision technology to realize three-dimensional reconstruction of hazardous chemicals warehouse is an effective means of safety monitoring in the warehouse,and stereo matching is the key and difficult point.A dense matching algorithm of binocular stereo pairs based on weighted SAD algorithm and gray level co-occurrence matrix is proposed.The algorithm improved the traditional SAD matching algorithm.The initial disparity map was obtained by introducing the weighted coefficient obeying the two-dimensional Gaussian distribution.Combined with the four texture features of the gray level co-occurrence matrix,the holes were filled with the image texture information,and the optimized disparity result was obtained.The experimental results show that the algorithm has good applicability in standard data set and simulation warehouse.

关 键 词：危化品仓储 双目立体视觉 立体匹配 SAD 灰度共生矩阵 

分 类 号：TP399[自动化与计算机技术—计算机应用技术] TP3-05[自动化与计算机技术—计算机科学与技术]