Exploring Polarimetric Properties Preservation during Reconstruction of PolSAR images using Complex-valued Convolutional Neural Networks

TL;DR

This paper addresses the challenge of preserving polarimetric information in PolSAR data during reconstruction by employing complex-valued Convolutional AutoEncoders. The authors leverage the intrinsic complex nature of PolSAR signals to maintain phase relationships and evaluate preservation through polarimetric decompositions, including Pauli, Krogager, Cameron, and $H-\alpha$ decompositions. The study provides extensive experiments on the San Francisco ALOS-2 and Brétigny datasets, showing that CV CoAEs achieve lower reconstruction error and superior preservation of polarimetric semantics compared to real-valued baselines, with favorable ablation results for activation, depth, and down/up-sampling choices. The work demonstrates the potential of physics-informed, complex-valued generative models for SAR data processing and sets a foundation for further exploration of complex-domain architectures in remote sensing.

Abstract

The inherently complex-valued nature of Polarimetric SAR data necessitates using specialized algorithms capable of directly processing complex-valued representations. However, this aspect remains underexplored in the deep learning community, with many studies opting to convert complex signals into the real domain before applying conventional real-valued models. In this work, we leverage complex-valued neural networks and investigate the performance of complex-valued Convolutional AutoEncoders. We show that these networks can effectively compress and reconstruct fully polarimetric SAR data while preserving essential physical characteristics, as demonstrated through Pauli, Krogager, and Cameron coherent decompositions, as well as the non-coherent $H-α$ decomposition. Finally, we highlight the advantages of complex-valued neural networks over their real-valued counterparts. These insights pave the way for developing robust, physics-informed, complex-valued generative models for SAR data processing.

Figures (15)

• Figure 1: $H-\alpha$ plane separated into areas 1 to 9, each corresponding to a specific scattering mechanism, with the entropy at the x-axis and the scattering angle at the y-axis. The black line represents the boundary of physically possible $H-\alpha$ couples.
• Figure 2: Architecture of a Convolutional AutoEncoder with the complex-valued residual blocks (yellow), and the complex-valued up-sampling layers (blue).
• Figure 3: Architecture of a Convolutional AutoEncoder with an explicit Bottleneck. The same color code as in Figure \ref{['fig:coae']} applies with the additional dense layers between the encoder and decoder.
• Figure 4: Distribution of the pixel-wise reconstruction error of the amplitude (left) (with a CVNN (left) and with an RVNN (right)) and phase (right) (with a CVNN (left) and with an RVNN (right)) between the original and reconstructed images obtained on the San Francisco Polarimetric SAR ALOS-2 dataset, with reconstruction error on the $x$-axis and number of pixels on $y$-axis.
• Figure 5: Distribution of the pixel-wise reconstruction error of the amplitude (left) (with a CVNN (left) and with an RVNN (right)) and phase (right) (with a CVNN (left) and with an RVNN (right)) between the original and reconstructed images obtained on the Brétigny dataset, with reconstruction error on the $x$-axis and number of pixels on $y$-axis.