# DTU-Net: A Multi-Scale Dilated Transformer Network for Nonlinear Hyperspectral Unmixing

arXiv: https://arxiv.org/abs/2503.03465 · HTML: https://mlsanity.com/paper/2503.03465/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 62 references that are in the map.
- First posted on arXiv 5 Mar 2025.

## Builds on (most cited first, 18 of 18)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [CBAM: Convolutional Block Attention Module](https://mlsanity.com/paper/1807.06521/) (2018)
- [Understanding the Behaviour of Contrastive Loss](https://mlsanity.com/paper/2012.09740/) (2020)
- [Early Convolutions Help Transformers See Better](https://mlsanity.com/paper/2106.14881/) (2021)
- [Spectral Variability in Hyperspectral Data Unmixing: A Comprehensive Review](https://mlsanity.com/paper/2001.07307/) (2020)
- [DilateFormer: Multi-Scale Dilated Transformer for Visual Recognition](https://mlsanity.com/paper/2302.01791/) (2023)
- [Nonlinear unmixing of hyperspectral images: models and algorithms](https://mlsanity.com/paper/1304.1875/) (2013)
- [Deep Hyperspectral Unmixing using Transformer Network](https://mlsanity.com/paper/2203.17076/) (2022)
- [Hyperspectral unmixing with spectral variability using a perturbed linear mixing model](https://mlsanity.com/paper/1502.01260/) (2015)
- [Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization](https://mlsanity.com/paper/1401.5649/) (2014)
- [Hyperspectral Unmixing via Deep Autoencoder Networks for a Generalized Linear-Mixture/Nonlinear-Fluctuation Model](https://mlsanity.com/paper/1904.13017/) (2019)
- [Integration of Physics-Based and Data-Driven Models for Hyperspectral Image Unmixing](https://mlsanity.com/paper/2206.05508/) (2022)
- [Nonparametric Detection of Nonlinearly Mixed Pixels and Endmember Estimation in Hyperspectral Images](https://mlsanity.com/paper/1503.05521/) (2015)
- [Unsupervised Nonlinear Spectral Unmixing based on a Multilinear Mixing Model](https://mlsanity.com/paper/1604.04293/) (2016)
- [Bi-Objective Nonnegative Matrix Factorization: Linear Versus Kernel-Based Models](https://mlsanity.com/paper/1501.05684/) (2015)
- [Estimating the Intrinsic Dimension of Hyperspectral Images Using an Eigen-Gap Approach](https://mlsanity.com/paper/1501.05552/) (2015)

## Built on by (most cited first, 1 of 1)

- [A Deep Equilibrium Network for Hyperspectral Unmixing](https://mlsanity.com/paper/2604.11279/) (2026)

## Abstract (arXiv)

Transformers have shown significant success in hyperspectral unmixing (HU). However, challenges remain. While multi-scale and long-range spatial correlations are essential in unmixing tasks, current Transformer-based unmixing networks, built on Vision Transformer (ViT) or Swin-Transformer, struggle to capture them effectively. Additionally, current Transformer-based unmixing networks rely on the linear mixing model, which lacks the flexibility to accommodate scenarios where nonlinear effects are significant. To address these limitations, we propose a multi-scale Dilated Transformer-based unmixing network for nonlinear HU (DTU-Net). The encoder employs two branches. The first one performs multi-scale spatial feature extraction using Multi-Scale Dilated Attention (MSDA) in the Dilated Transformer, which varies dilation rates across attention heads to capture long-range and multi-scale spatial correlations. The second one performs spectral feature extraction utilizing 3D-CNNs with channel attention. The outputs from both branches are then fused to integrate multi-scale spatial and spectral information, which is subsequently transformed to estimate the abundances. The decoder is designed to accommodate both linear and nonlinear mixing scenarios. Its interpretability is enhanced by explicitly modeling the relationships between endmembers, abundances, and nonlinear coefficients in accordance with the polynomial post-nonlinear mixing model (PPNMM). Experiments on synthetic and real datasets validate the effectiveness of the proposed DTU-Net compared to PPNMM-derived methods and several advanced unmixing networks.