基于单目视觉的目标定位算法研究

Research of Target Positioning Algorithm Based on Monocular Vision

针对先进制造系统中的目标实时定位问题,提出了一种基于单目视觉的目标定位方法.首先,通过构建相机模型,建立图像像素坐标与世界坐标之间的关系,并通过相机标定获得相机内参数;然后,采用Harris算法提取并匹配图像的特征点,并使用RANSCA算法根据匹配的特征点对来计算图像的基础矩阵,从而推导出图像的本质矩阵,获得相机外参数;在此基础上,计算出目标的精确坐标;最后,通过一组实验验证所提方法的有效性和可行性.

The method of target positioning based on monocular vision for target real-time positioning problem in advanced manufacturing system is proposed.Firstly,by building the camera model, the relationship between the image pixel coordinate and world coordinate is established,and the camera intrinsic parameters are obtained with the camera calibration; Then, the feature points of the images are extracted and matched with the Harris algorithm, and the fundamental matrix of the images are computed according to the matched of feature points with the RANSAC algorithm,thus the essential matrix of the images is deduced, and the external parameters of the camera are obtained;on this basis, the precise coordinates of target are deduced.Finally,the feasibility of the method in this paper is verified by a set of experlments.