Digital Elevation Model enhancement using Deep Learning

TL;DR

This paper presents a deep learning method to enhance planetary digital elevation models, achieving significant resolution improvements and robust feature recovery under various lighting conditions, with automated processing at a global scale.

Contribution

It introduces a novel deep learning approach for DEM enhancement that surpasses traditional methods in resolution and robustness, enabling automated global-scale processing.

Findings

Achieves 90x resolution improvement in Mars DEMs

Recovers features obscured by lighting conditions

Enhanced slope errors comparable to high-resolution maps

Abstract

We demonstrate high fidelity enhancement of planetary digital elevation models (DEMs) using optical images and deep learning with convolutional neural networks. Enhancement can be applied recursively to the limit of available optical data, representing a 90x resolution improvement in global Mars DEMs. Deep learning-based photoclinometry robustly recovers features obscured by non-ideal lighting conditions. Method can be automated at global scale. Analysis shows enhanced DEM slope errors are comparable with high resolution maps using conventional, labor intensive methods.