Deep learning method reduces date-to-date bias in satellite images
Monotone Piecewise-Linear Harmonization of Multitemporal Multispectral Surface Reflectance With Unsupervised Deep Multiview Pseudoinvariant Features Learning
Computer Vision and Pattern Recognition
Summary
Satellite images taken at different times sometimes show small color and brightness differences, even when the ground hasn't changed. This study introduces a new deep learning approach that finds stable features across multiple images to fix these biases without needing any labeled data. The method uses advanced mathematical techniques to better align images from different dates, improving the consistency of satellite data. Tests on several years of satellite images show this approach reduces errors and keeps the true changes on the ground intact.
What this means in practice
- •For remote sensing engineers: Improve multitemporal satellite image alignment by reducing date-specific biases in surface reflectance data.
- •For agriculture monitoring teams: Use corrected satellite reflectance data to get more reliable information about crop health and changes over time.
Authors
Lino Garda Denaro, Chia-Hsiang Lin
Abstract
Surface reflectance products derived from optical satellite may exhibit systematic inconsistencies across acquisition dates, because atmospheric correction is generally performed independently for each image. Such residual biases distort the temporal behavior of invariant ground targets and reduce the reliability of multitemporal surface reflectance analysis. This study proposes Deep Multitemporal Generalized Canonical Correlation Analysis (DeepMGCCA), an unsupervised deep multiview pseudoinvariant feature learning framework that integrates Deep Generalized Canonical Correlation Analysis (GCCA) with Multivariate Alteration Detection (MAD) to identify deep-pseudoinvariant features (DIFs) for the removal of surface reflectance biases across multitemporal satellite observations. Unlike conventional MAD-based approaches that rely on linear GCCA, the proposed framework replaces date-specific linear projections with a shared convolutional encoder. The framework learns deep invariant features (DIFs) directly from pixel-weighted canonical variates via iterative reweighting strategy, while a reconstruction regularizer preserves surface reflectance information and prevents degenerate collapse from suppressing genuine temporal variability. Experiments were conducted on four annual Landsat-8/9 and Sentinel-2 acquisitions across four regions of interest (ROIs) to evaluate four DIF-based harmonization transforms. A monotone piecewise-linear transform derived from DIFs quantiles consistently achieved the largest bias reduction, lowering the between-date spectral angle by up to 2.54 and RMSE by 0.015, and raising PSNR by up to 4.04~dB. These findings demonstrate the potential of deep invariant feature learning for multitemporal surface reflectance bias removal while highlighting the importance of preserving physically meaningful temporal and spectral information during nonlinear representation learning.