Lidar Upsampling with Sliced Wasserstein Distance

TL;DR

This paper introduces a novel lidar upsampling method that leverages edge-aware dense convolutions and Sliced Wasserstein Distance to reconstruct fine-grained lidar scan patterns, improving over existing techniques.

Contribution

The proposed approach combines edge-aware dense convolutions with Sliced Wasserstein Distance for effective lidar point cloud upsampling without coarse-fine stages.

Findings

Achieves better upsampling quality than previous methods.

Effectively reconstructs fine-grained lidar scan patterns.

Operates in a one-stage upsampling framework.

Abstract

Lidar became an important component of the perception systems in autonomous driving. But challenges of training data acquisition and annotation made emphasized the role of the sensor to sensor domain adaptation. In this work, we address the problem of lidar upsampling. Learning on lidar point clouds is rather a challenging task due to their irregular and sparse structure. Here we propose a method for lidar point cloud upsampling which can reconstruct fine-grained lidar scan patterns. The key idea is to utilize edge-aware dense convolutions for both feature extraction and feature expansion. Additionally applying a more accurate Sliced Wasserstein Distance facilitates learning of the fine lidar sweep structures. This in turn enables our method to employ a one-stage upsampling paradigm without the need for coarse and fine reconstruction. We conduct several experiments to evaluate our method and demonstrate that it provides better upsampling.