双目视觉无人机图像测量系统设计与实现  

Design and Implementation of Binocular Vision UAV Image Measurement System

作　　者：王浩同 刘智平[1] 石俊 刘白林[1] 王鹏瑞 WANG Haotong;LIU Zhiping;SHI Jun;LIU Bailin;WANG Pengrui(School of Computer Science and Engineering,Xi’an Technological University,Xi’an 710021,China)

机构地区：[1]西安工业大学计算机科学与工程学院,西安710021

出　　处：《西安工业大学学报》2020年第5期541-548,共8页Journal of Xi’an Technological University

基　　金：陕西省工业领域重点项目(2016KTZDGY4-09)。

摘　　要：为了解决“低慢小”目标低空检测效率低与空间定位难的问题,文中结合双目立体视觉深度测量的优点,设计并实现了一种无人机图像测量系统。用三帧差法检测低空非合作目标,根据几何不变矩原理实现目标轮廓特征匹配识别。通过建立双目视觉测量数学模型,利用最小二乘法求解目标空间三维坐标,并建立实验分析测量误差。实验结果表明:保证双目相机帧率至少在15 fps前提下,通过多线程改进匹配算法,提高系统效率15%左右;同时在双目相机20 m深度感知范围内,目标空间三维坐标测量误差均值近0.2 m,满足系统实时性和精度的要求。With the advantages of the binocular stereo vision depth measurement,a UAV image measurement system is designed and implemented for the low-altitude detection and spatial positioning of“low,slow and small”targets.The three-frame difference method is used to detect low-altitude non-cooperative targets,and the target contour feature matching recognition is realized according to the principle of geometric invariant moments.A mathematical model of the binocular vision measurement is built,and the least square method is used to solve the three-dimensional coordinates of the target space.An experiment is conducted to analyze the measurement errors.Experimental results show that under the premise that the frame rate of the binocular camera is at least 15 fps,the matching algorithm is improved by multi-threading and the system efficiency is improved by 15%or so.And,within the 20 m depth perception range of the binocular camera,the average error of 3D coordinate measurement of the target space is close to 0.2 m,which meets what the real-time and accuracy of the system require.

关 键 词：无人机 三帧差法 模板匹配 双目视觉测量 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]