Infrared Colorization Using Deep Convolutional Neural Networks

TL;DR

This paper introduces a deep learning method for converting near-infrared images to natural-looking RGB images, preserving details without needing user guidance or reference images, trained on real-world road scene data.

Contribution

It presents a novel deep multi-scale convolutional neural network approach for NIR to RGB colorization that does not require user input or reference images during inference.

Findings

The method produces natural-looking RGB images from NIR inputs.

It effectively preserves high-frequency details of the original NIR images.

The approach is validated on a large real-world dataset of road scenes.

Abstract

This paper proposes a method for transferring the RGB color spectrum to near-infrared (NIR) images using deep multi-scale convolutional neural networks. A direct and integrated transfer between NIR and RGB pixels is trained. The trained model does not require any user guidance or a reference image database in the recall phase to produce images with a natural appearance. To preserve the rich details of the NIR image, its high frequency features are transferred to the estimated RGB image. The presented approach is trained and evaluated on a real-world dataset containing a large amount of road scene images in summer. The dataset was captured by a multi-CCD NIR/RGB camera, which ensures a perfect pixel to pixel registration.