# Laplacian reconstructive network for guided thermal super-resolution

## TL;DR

This paper introduces LapGSR, a lightweight model that improves thermal image resolution using RGB data and Laplacian pyramids, achieving high performance with fewer parameters.

## Contribution

The novel LapGSR model uses Laplacian pyramids and a combined loss to achieve efficient, high-quality thermal super-resolution with fewer parameters.

## Key findings

- LapGSR achieves excellent results on the ULB17-VT and VGTSR datasets.
- The model has significantly fewer parameters than state-of-the-art models.
- It preserves spatial and structural details while being computationally efficient.

## Abstract

In the last few years, the fusion of multi-modal data has been widely studied for various applications such as robotics, gesture recognition, and autonomous navigation. Indeed, high-quality visual sensors are expensive, and consumer-grade sensors produce low-resolution images. Researchers have developed methods to combine RGB colour images with non-visual data, such as thermal, to overcome this limitation to improve resolution. Fusing multiple modalities to produce visually appealing, high-resolution images often requires dense models with millions of parameters and a heavy computational load, which is commonly attributed to the intricate architecture of the model. We propose LapGSR, a multimodal, lightweight, generative model incorporating Laplacian image pyramids for guided thermal super-resolution. This approach uses a Laplacian Pyramid on RGB colour images to extract vital edge information, which is then used to bypass heavy feature map computation in the higher layers of the model in tandem with a combined pixel and adversarial loss. LapGSR preserves the spatial and structural details of the image while also being efficient and compact. This results in a model with significantly fewer parameters than other SOTA models while demonstrating excellent results on two cross-domain datasets viz. ULB17-VT and VGTSR datasets.