GAUSS: Guided encoder - decoder Architecture for hyperspectral Unmixing with Spatial Smoothness

• Published in: European journal of remote sensing Vol. 56; no. 1
• Main Authors: Wickramathilaka, H.M.K.D., Fernando, D., Jayasundara, D., Wickramasinghe, D., Ranasinghe, D.Y.L., Godaliyadda, G.M.R.I., Ekanayake, M.P.B., Herath, H.M.V.R., Ramanayake, L., Senarath, N., Weerasooriya, H.M.H.K.
• Format: Journal Article
• Language: English
• Published: Cagiari Taylor & Francis Ltd 31.12.2023; Taylor & Francis Group
• Subjects: Algorithms; Architecture; autoencoder; Constraint modelling; Datasets; Decomposition; Deep learning; Encoders-Decoders; Hyperspectral unmixing; Machine learning; Remote sensing; Signal processing; Smoothness; Sparsity; spatial smoothness and correlation; split architecture; supervised learning; Training
• ISSN: 2279-7254; 2279-7254
• DOI: 10.1080/22797254.2023.2277213

This study introduces GAUSS (Guided encoder-decoder Architecture for hyperspectral Unmixing with Spatial Smoothness), a novel autoencoder-based architecture for hyperspectral unmixing (HU). GAUSS consists of an Approximation Network (AN), Unmixing Network (UN), and a Mixing Network (MN). The AN incorporates spatial context within a hyperspectral pixel’s neighborhood, while the UN utilizes a pseudo-ground truth mechanism to enhance abundance estimation. The MN provides estimated endmembers’ signatures. By incorporating UN-produced abundances, unlike the conventional AE model, GAUSS overcomes the single-layer constraint of the MN. Thereafter, a secondary training phase improves the accuracy of endmembers and abundance estimation using a reliable Signal Processing (SP) algorithm, resulting in superior HU performance. The results demonstrate the effectiveness of GAUSS on two Standard datasets and a Simulated dataset compared to the state-of-the-art SP and Deep Learning (DL) based methods. This signifies the benefit of integrating an SP algorithm in the training process, contributing to advancements in DL-based HU techniques.