无人机地面目标实时自主视觉定位  

Ground targets location of UAV using vision location method

作　　者：王向军[1,2] 万子敬 王鑫[1] 刘世廷[1] 文鹏程[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072 [2]天津大学微光机电系统技术教育部重点实验室,天津300072

出　　处：《红外与激光工程》2014年第2期615-619,共5页Infrared and Laser Engineering

基　　金：国家教育部支撑项目(625010110);国家自然科学基金与中国民用航空总局联合研究基金(61179043)

摘　　要：微小型无人机携带设备受到重量限制,基于摄像机透视投影原理的坐标解算模型,重点研究基于数字高程模型的测高逼近算法,获取无人机距离地面目标之间的高度差,以满足微小型无人机机载设备实时自主定位地面目标的要求。计算机仿真实验结果,以及完成算法硬件平台移植后进行的无人机实飞测试实验结果均可以看出,采用的解算模型结合基于数字高程模型的测高逼近算法可以满足对25帧/s视频流的地面目标实时自主定位,并具有较高的精度。The equipment of Micro Unmanned Aerial Vehicle （MUAV） was limited by weight. In this paper, the coordinate solver model based on machine vision was introduced, and on this basis, the iterative approximation algorithm which was based on Digital Elevation Model （DEM） was considered to get the altitude difference between the Unmanned Aerial Vehicle （UAV） and the target in order to locating the ground target caught by the optronics pod with the airborne equipment of an MUAV autonomously. The computer simulations and the flight experimental results from program which was transplanted to the airborne devices showed that the coordinate solver model combined with the iterative approximation algorithm could calculate the position of ground target in the video stream （25fps） timely and accurate.

关 键 词：微小型无人机 数字高程模型 机器视觉 目标定位 

分 类 号：V243.5[航空宇航科学与技术—飞行器设计] TP391.4[自动化与计算机技术—计算机应用技术]