摘要
为了实现对近景目标的高速三维跟踪,设计了一种小型仿生复眼系统,并将其与可开窗相机相结合。介绍了系统的结构与设计原则,提出了一种适用于多孔径系统的跟踪与探测策略。利用卡尔曼滤波器对图像传感器进行多路并行开窗,从而克服视场与空间分辨率对跟踪速度的制约,在较低代价的前提下实现了高速三维跟踪。采用蔡氏两步法对7子眼复眼镜头进行了统一标定,并建立了归一化多目探测模型,在实时跟踪的基础上实现了三维信息获取。为了验证系统的高速三维跟踪效果,在3 000f/s的帧频下,对一个以10r/s的角速度做匀速圆周运动的目标进行了实时跟踪,并重构出目标在距相机150mm处的三维运动轨迹。实验结果表明,系统能够有效实现对高速运动目标的实时跟踪与三维粗定位。分析了影响系统三维探测精度的因素以及提高精度的方法,为用于全景高速三维探测的复眼跟踪系统的研制奠定了基础。
A compact compound-eye model is designed to track fast-moving objects and detect their three dimensional（3D） coordinates,which is combined with a camera with dynamic regions of interest （ROD. The structure of this system and the design principles are presented,and a strategy for tracking and detecting,which is suitable for compound-eye system,is proposed. Kalman filters are employed to parallelly generate multiple regions of interest on the imaging sensor, thus resolving the conflicts among wide field of vision,high resolution and real time characteristic and realizing high-speed tracking at a low cost. The 7 ommatidia of the compound-eye lens are calibrated using Tsai two-step method, and a normalized multi-view detecting model is established to obtain the 3D information of fast-moving objects. This system is used to track an object with uniform circular motion ata speed of 10 r/s at a frame rate of 3000 f/s,and obtain the 3D trajectory of the object at a distance of 150 ram. Preliminary experimental results show that the compound-eye system can effectively track and roughly locate high-speed objects in real time. Factors affecting the detecting precision are analyzed, and effective measures to improve the accuracy are discussed,which might lay the foundation for the future design of panoramic high-speed 3D-track- ing compoundeye system.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2014年第9期1796-1802,共7页
Journal of Optoelectronics·Laser
基金
国家重点基础研究发展计划(2011CB706706)
国家外专局引智计划(B07014)
国家自然科学基金(51175377)
天津市自然科学基金(12JCQNJC02700)资助项目
关键词
复眼
开窗
卡尔曼滤波
高速
三维跟踪
compound-eye
regions of interest
Kalman filter
high-speedl 3D tracking
