基于合作目标的仪表盘自适应虚实配准方法  

An Adaptive Virtual-Real Registration Method for Instrument Panel Based on Cooperative Target

作　　者：李浩铭 付战平 胡文婷 王曼曼 冯亚杰 LI Haoming;FU Zhanping;HU Wenting;WANG Manman;FENG Yajie(Naval Aviation University Qingdao Campus,Qingdao 266000 China;Naval Aviation University College of Coastal Defense,Yantai 264000 China;No.7 Middle School Yongji County,Jilin 132000 China)

机构地区：[1]海军航空大学青岛校区,山东青岛266000 [2]海军航空大学岸防兵学院,山东烟台264000 [3]永吉县第七中学,吉林吉林132000

出　　处：《电光与控制》2022年第10期58-63,共6页Electronics Optics & Control

基　　金：海军航空大学科研基金(I32004008)。

摘　　要：在模拟器虚实融合实验中,虚拟坐标与真实坐标配准是一项非常重要的工作。特别在飞行模拟器仪表按钮较为繁杂的情况下,精准识别仪表盘比较耗时,难以满足摄像头环境下仪表盘实时定位的要求。在视频透视头盔已实现鱼眼图像矫正和获得深度图像的基础上,提出了一种基于合作目标的仪表盘自适应虚实配准方法。通过手工标定帧数据中的合作目标中心坐标数据,计算出6种传统模板匹配算法的识别误差及算法处理时间,选出了误差最小的算法。进一步通过增加自适应匹配功能改进了识别算法,大大提高了视频流识别的准确率。通过坐标变换得到了仪表盘真实空间坐标,为UE4场景中的虚拟仪表盘进行三维坐标校准做铺垫,实现了更好的虚实配准效果。The registration of virtual coordinate and real coordinate is a very important task in the virtual-real fusion experiment of simulators.Especially in the case of complicated buttons of flight simulator instrument it is time-consuming to accurately identify instrument panel which is difficult to meet the requirements of real-time positioning of instrument panel in camera environment.Based on the realization of fish-eye image correction and depth image acquisition by video see-through helmet an adaptive virtual-real registration method for instrument panel based on cooperative target is proposed.The identification error and processing time of six kinds of traditional template matching algorithms are calculated by using the coordinate data of cooperative target center in manual calibration frame data and the algorithm with the least error is selected.Furthermore the recognition algorithm is improved by adding adaptive matching function which greatly improves video stream recognition accuracy.Through coordinate transformation the real spatial coordinates of instrument panel are obtained which paves the way for 3D coordinate calibration of virtual instrument panel in Unreal Engine 4(UE4)scene and achieves better virtual-real registration effect.

关 键 词：视频流识别 合作目标 自适应匹配 虚实配准 UE4 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]