Code-Aligned Autoencoders for Unsupervised Change Detection in Multimodal Remote Sensing Images

Image translation with convolutional autoencoders has recently been used as an approach to multimodal change detection in bitemporal satellite images. A main challenge is the alignment of the code spaces by reducing the contribution of change pixels to the learning of the translation function. Many...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2020-04
Main Authors: Luppino, Luigi T, Hansen, Mads A, Kampffmeyer, Michael, Bianchi, Filippo M, Moser, Gabriele, Jenssen, Robert, Anfinsen, Stian N
Format: Article
Language:English
Subjects:
Affinity
Algorithms
Alignment
Change detection
Image detection
Machine learning
Pixels
Remote sensing
Satellite imagery
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Image translation with convolutional autoencoders has recently been used as an approach to multimodal change detection in bitemporal satellite images. A main challenge is the alignment of the code spaces by reducing the contribution of change pixels to the learning of the translation function. Many existing approaches train the networks by exploiting supervised information of the change areas, which, however, is not always available. We propose to extract relational pixel information captured by domain-specific affinity matrices at the input and use this to enforce alignment of the code spaces and reduce the impact of change pixels on the learning objective. A change prior is derived in an unsupervised fashion from pixel pair affinities that are comparable across domains. To achieve code space alignment we enforce that pixel with similar affinity relations in the input domains should be correlated also in code space. We demonstrate the utility of this procedure in combination with cycle consistency. The proposed approach are compared with state-of-the-art deep learning algorithms. Experiments conducted on four real datasets show the effectiveness of our methodology.
ISSN:2331-8422