Precise and Robust Color Point Cloud Registration with Correntropy-Based ICP Algorithm

The paper proposes a variant of the Iterative Closest Point (ICP) algorithm for point cloud registration that addresses two main issues with existing ICP-based algorithms: the lack of precise and complete color information and the lack of robustness to noise and outliers. The proposed algorithm introduces a color distance metric based on the L*a*b* color space and a correntropy-based objective function to enhance the accuracy and robustness of the algorithm. Simulation experiments on an RGB-D object dataset show that the proposed algorithm outperforms existing ICP-based methods in terms of accuracy and robustness.

I. Introduction

Point cloud registration is a transformation prediction problem to align two point clouds collected from the same object or scene. Point cloud registration has been widely used in many computer vision fields, such as 3D reconstruction, pose estimation, and self-localisation and mapping (SLAM) [1]. With the development of 3D sensors such as LiDAR and RGB-D cameras, point cloud registration has attracted more and more attention and many research works have been proposed in this field.