论文部分内容阅读
为使机器人同时具备双目立体视觉和单目运动视觉的仿人化环境感知能力,克服双目视场狭窄、单目深度感知精度低的缺陷,本文基于人眼结构特点,设计了一个具有4个旋转自由度的双目仿生眼平台,并分别基于视觉对准策略和手眼标定技术实现了该平台的初始定位和参数标定.给出了基于外部参数动态变化的双目立体感知方法和单目运动立体感知方法,前者通过两架摄像机实时获取的图像信息以及摄像机相对位姿信息进行3维感知,后者综合利用单个摄像机在多个相邻时刻获取的多个图像及其对应姿态进行3维感知.实验结果中的双目视觉相对感知精度为0.38%,单目运动视觉相对感知精度为0.82%.本文方法不但能够有效拓宽传统双目视觉的感知视野,而且能够保证双目感知和单目运动感知的准确性.
In order to overcome the shortcomings of binocular field of view and monocular depth perception, the robot has the human-like environment perception capability of binocular stereo vision and monocular sports vision, and based on the structural characteristics of the human eye, a four- A rotational degree of freedom binocular biometric eye platform, and the initial positioning and parameter calibration of the platform are realized respectively based on the visual alignment strategy and the hand-eye calibration technology.An approach to binocular stereo perception based on the dynamic change of external parameters and the monocular Motion perception method, the former three-dimensional perception of the image information acquired by the two cameras in real time and the relative position and orientation information of the camera, the latter three-dimensional using a plurality of images acquired by a single camera at multiple adjacent moments and their corresponding pose The results show that the relative perceived accuracy of binocular vision is 0.38%, and the relative perceived accuracy of monocular motor vision is 0.82% .This method not only can effectively broaden the traditional visual field of binocular vision, but also can ensure binocular perception and monocular Accuracy of motion perception.