基于目标运动的复眼式双目视觉测距法

昆虫（螳螂）复眼利用目标的动态差异实现立体视觉，具有视场大、计算简单、实时性高等特点，是立体视觉研究的新方向。为实现复眼立体视觉在机器人视觉导航中的应用，根据复眼结构以及信息处理机制，提出并研究了环形光电传感器的目标快速检测与定位方法。首先，搭建了基于60个光电二极管的等6°夹角分布式环形传感器，形成具有360°视场的环形仿生复眼；其次，建立了基于光流原理的运动目标方位角检测模型，采用傅里叶拟合法实现了方位角检测模型的优化，实现了运动目标方位角与目标距离大视野范围内的简单、快速检测。实验结果表明:1）可实现距离375 mm范围内运动速度为30 mm/s的目标方位角实时检测，测量误差在2°范围内；2）基于目标方位角检测模型，可实现双目阵列传感器在300 mm×375 mm视野重叠区域范围内、平均测量误差在10 mm范围内的立体视觉测距。基于光流的运动目标方位角检测模型可用于实现对运动目标空间位置的动态实时检测，在运动检测、视觉导航等领域具有广泛的应用前景。

Compound eye binocular vision ranging method based on target motion

The insect (mantis) compound eye realizes the stereo vision by using dynamic difference of the target, which has the characteristics of large field of view, simple calculation and high real-time performance, and is a new direction of the stereo vision research. In order to realize the application of compound eye stereo vision in robot vision navigation, according to the compound eye structure and the information processing mechanism, a target rapid detection and location method based on ring photoelectric sensor was proposed and studied. Firstly, a ring sensor based on 6° angle distribution of 60 photodiodes was built to form an ring bionic compound eye with 360° field of view. Secondly, a rapid detection model of the moving target azimuth based on optical flow principle was established. The Fourier fitting method was adapted to optimize the azimuth detection model, and the simple and rapid detection in the range of the large field of view of the moving target azimuth and the target distance was realized. The experimental results show that: 1) the real-time detection of the target azimuth at the motion velocity of 30 mm/s in the range of 375 mm can be realized, and the measurement error is within 2°; 2) based on the target azimuth detection model, the stereo vision ranging of the binocular array sensor in the range of 300 mm×375 mm field of view overlap with an average measurement error of 10 mm can be realized. The moving target azimuth detection model based on optical flow can be used to realize the dynamic real-time detection of the moving target spatial position, and has broad application prospects in the fields of motion detection and vision navigation.