基于形状不变特征和三维几何参数的工件识别方法

Recognizing Industrial Parts Using 2D Shape Invariant Features and 3D Geometrical Parameters

该文提出一种扁平工件识别新方法。首先根据图象平面和工作台平面之间的变换计算出工件轮廓点的近似三维坐标,并根据这些近似三维坐标计算出有关的几何参数,然后利用几何参数和工件物体的形状不变特征,采用分层匹配的方法进行识别。根据文中方法设计并实现了一个由单个摄像机构成的工件识别系统,实验结果表明该方法是可行且有效的。

In this paper, a new method is proposed for recognitionof flat parts. First, the 3D coordinates of the points of parts are calcu-lated approximately by the transform between the table plane and im-age plane, and the 3D geometrical parameters of the parts are calculat-ed based on the approximate coordinates. Then, these geometrical pa-rameters are combined with 2D shape invariant features of parts forrecognition. A machine vision system for recognizing industrial partson the basis of the presented method has been developed. The experi-mental results showed the system works well for recognizing shapedistorted parts.