The camera calibration, feature extraction and stereo matching are key links in the vision system of excavator robot, especially the stereo matching which is the most complex and important step in the vision system. This paper analyzes the general and parallel stereo binocular visual models and studies the measurement method of parallel binocular stereo vision system and parameters calibration of the excavator robot. Combined with the matching method based on characteristic and according to the extracted angular point of left and right images, it realizes the matching for feature points based on epipolar geometry constrain. Through the experiment of image matching for bucket, the method is verified that it meets the requirements of the vision system of excavator robot.%在挖掘机器人视觉系统实现中,摄像机标定、特征提取、立体匹配是关键环节,而立体匹配又是视觉系统中最复杂、最重要的步骤.首先分析了通用双目立体视觉模型及平行双目立体视觉模型,对挖掘机器人平行双目立体视觉系统测量方法及参数标定进行了研究.结合基于特征的匹配方法,通过对左、右图像角点的提取,实现了基于极线几何约束的特征点匹配.通过对铲斗图像匹配实验,验证了该方法能满足挖掘机器人视觉系统要求.
展开▼