Research on Component Recognition and Attitude Estimation Method Based on Point Cloud

Abstract

In order to identify and understand the real objects in the assembly environment, propose a 3D aware method of registration based on real scene object perception by the similarity of point cloud, in order to reduce the impact of noise on the registration results, propose a method of matching algorithm based on adaptive density, the density of point cloud adaptive adjustment of different sources. Using the affine invariance of four coplanar point features to achieve recognition of real objects. This method can reduce the amount of cloud data, reduce the computational complexity, and also improve the matching accuracy. The experimental results show that the proposed method can identify components accurately in the assembly environment point cloud and calculate their pose, and can be used for augmented reality interactive applications under noisy conditions.