LiDAR depth completion is a task that predicts depth values for every pixel on the corresponding camera frame . Most of the existing state-of-the-art solutions are based on deepneural networks . In this letter, a novel non-learning depth completionmethod is proposed by exploiting the local surface geometry that is enhanced by an outlier removal algorithm . On KITTI dataset, the proposed solution achievesthe best error performance among all existing methods and is comparable to the best self-supervised learning method and some supervisedlearning methods . Moreover, since outlier points from occluded regions is a commonly existing problem, it is a generalpreprocessing step that is applicable to many robotic systems with both cameraand LiDar sensors . The proposed method is a . proposed algorithm is also a general pre-processing step to many robots with both cameras and LiDare sensors. The proposed outlier removed algorithm is a generic pre-pre-processing step that can be applied to many systems with LiDARE sensors. It is available on KITTi dataset, and it is available to many automated robots with LiAR sensors

Author(s) : Yiming Zhao, Lin Bai, Ziming Zhang, Xinming Huang

Links : PDF - Abstract

Code :

https://github.com/jiangzhongshi/SurfaceNetworks


Coursera

Keywords : proposed - depth - sensors - step - pre -

Leave a Reply

Your email address will not be published. Required fields are marked *