Guided thermal image super-resolution
Abstract
Deep learning changed the era of traditional computer vision by proposing algorithms that can achieve great success in computer vision compared with traditional methods in several areas such as image segmentation, object detection, object classification, image enhancement, speech recognition, image processing, text recognition, and among others. With the fast growth in the visual surveillance and security sectors, infrared images have become increasingly necessary in a large variety of applications. This is true even though IR sensors are still more expensive than their RGB counterpart having a considerably low-resolution. In general, images from the visible spectrum (e.g., RGB) have rich enough information, but sometimes objects can appear in different conditions of illumination, occlusion, and background clutter. These conditions can severely degrade the system’s performance. Therefore visible data is found to be insufficient and infrared images are becoming a standard tool to tackle these problems despite the fact of their poor resolution. This project aims at developing novel deep learning-based architectures that exploit the rich information from one spectral band to improve images from another spectral band — a cross-spectral framework. In this context, we will try to boost the infrared image resolution by using the corresponding high-resolution visible image as an auxiliary input to the model to exploit information of high-resolution representation from the visible spectrum to improve low-resolution infrared images.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 07, 2023
- Source ID
- FA95502210261
Entities
People
- Angel Sappa
Organizations
- Air Force Office of Scientific Research
- United States Air Force