Digital Elevation Model enhancement using Deep Learning
This work addresses the need for high-resolution terrain mapping in planetary science, offering an automated solution that improves upon labor-intensive techniques.
The paper tackles the problem of enhancing planetary digital elevation models (DEMs) by using optical images and deep learning, achieving a 90x resolution improvement in global Mars DEMs with slope errors comparable to conventional methods.
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.