融合MeanShift和改进SURF算法的目标定位策略

Targeting Strategy Based on MeanShift and Improved SURF Algorithm

在工业环境中,针对工业机器人进行抓取物品操作时定位耗时长,精度低的问题,提出一种融合MeanShift算法和改进SURF算法的分层式定位策略,快速精确地定位物体。首先根据双目视觉系统采集的目标物体图片信息,采用MeanShift算法进行初步处理,切割出目标图片信息;然后利用改进后SURF算法对目标区域进行特征点对的匹配以及筛选;最后将匹配好的特征点对根据三角形测量原理,实现物体三维坐标的精确定位。实验验证了在工业机器人抓取物品时,本文所提方法对物体定位速度与精度上有所提升。

In the industrial environment,a hierarchical object localization strategy combining Mean Shift algorithm and improved SURF algorithm is proposed to locate the object quickly and accurately when the object is manipulated by the industrial robot.Firstly,according to the target object image information collected by the binocular vision system,the MeanShift algorithm is used for cutting out the target image information for initial processing.Then,the feature points are matched by the improved SURF algorithm with good computational superiority and repeatability,uniqueness and robustness.Finally,the matched feature points are combined with triangular measurement principle to accurately locate the three-dimensional coordinates of the object.The experiment verifies that the method improves the speed and accuracy of object positioning when industrial robots grasp objects.