Physically Explainable CNN for SAR Image Classification

by   Zhongling Huang, et al.

Integrating the special electromagnetic characteristics of Synthetic Aperture Radar (SAR) in deep neural networks is essential in order to enhance the explainability and physics awareness of deep learning. In this paper, we firstly propose a novel physics guided and injected neural network for SAR image classification, which is mainly guided by explainable physics models and can be learned with very limited labeled data. The proposed framework comprises three parts: (1) generating physics guided signals using existing explainable models, (2) learning physics-aware features with physics guided network, and (3) injecting the physics-aware features adaptively to the conventional classification deep learning model for prediction. The prior knowledge, physical scattering characteristic of SAR in this paper, is injected into the deep neural network in the form of physics-aware features which is more conducive to understanding the semantic labels of SAR image patches. A hybrid Image-Physics SAR dataset format is proposed, and both Sentinel-1 and Gaofen-3 SAR data are taken for evaluation. The experimental results show that our proposed method substantially improve the classification performance compared with the counterpart data-driven CNN. Moreover, the guidance of explainable physics signals leads to explainability of physics-aware features and the physics consistency of features are also preserved in the predictions. We deem the proposed method would promote the development of physically explainable deep learning in SAR image interpretation field.


page 6

page 12

page 17


Explainable Analysis of Deep Learning Methods for SAR Image Classification

Deep learning methods exhibit outstanding performance in synthetic apert...

Explainable, Physics Aware, Trustworthy AI Paradigm Shift for Synthetic Aperture Radar

The recognition or understanding of the scenes observed with a SAR syste...

Physics Guided Deep Learning for Data-Driven Aircraft Fuel Consumption Modeling

This paper presents a physics-guided deep neural network framework to es...

Explainability in CNN Models By Means of Z-Scores

This paper explores the similarities of output layers in Neural Networks...

Guided Unsupervised Learning by Subaperture Decomposition for Ocean SAR Image Retrieval

Spaceborne synthetic aperture radar (SAR) can provide accurate images of...

Recognition of polar lows in Sentinel-1 SAR images with deep learning

In this paper, we explore the possibility of detecting polar lows in C-b...

Real-time simulation of viscoelastic tissue behavior with physics-guided deep learning

Finite element methods (FEM) are popular approaches for simulation of so...

Please sign up or login with your details

Forgot password? Click here to reset