Point Cloud Color Constancy

TL;DR

This paper introduces PCCC, a fast and effective point cloud-based algorithm for estimating scene illumination chromaticity using RGB-D data and PointNet architecture, outperforming existing methods.

Contribution

The paper presents a novel point cloud approach for color constancy leveraging depth information and PointNet, achieving higher accuracy and speed.

Findings

Lower error than state-of-the-art algorithms on RGB-D datasets.

Achieves over 500 fps in processing speed.

Effective use of depth and RGB data for illumination estimation.

Abstract

In this paper, we present Point Cloud Color Constancy, in short PCCC, an illumination chromaticity estimation algorithm exploiting a point cloud. We leverage the depth information captured by the time-of-flight (ToF) sensor mounted rigidly with the RGB sensor, and form a 6D cloud where each point contains the coordinates and RGB intensities, noted as (x,y,z,r,g,b). PCCC applies the PointNet architecture to the color constancy problem, deriving the illumination vector point-wise and then making a global decision about the global illumination chromaticity. On two popular RGB-D datasets, which we extend with illumination information, as well as on a novel benchmark, PCCC obtains lower error than the state-of-the-art algorithms. Our method is simple and fast, requiring merely 16*16-size input and reaching speed over 500 fps, including the cost of building the point cloud and net inference.