An Augmented Linear Mixing Model to Address Spectral Variability for Hyperspectral Unmixing
By Danfeng Hong, Naoto Yokoya, Jocelyn Chanussot, Xiaoxiang Zhu. 2018
828 citations recorded by the source corpus.
Hyperspectral imagery collected from airborne or satellite sources inevitably suffers from spectral variability, making it difficult for spectral unmixing to accurately estimate abundance maps. The classical unmixing model, the linear mixing model (LMM), generally fails to handle this sticky issue effectively. To this end, we propose a novel spectral mixture model, called the augmented LMM, to address spectral variability by applying a data-driven learning strategy in inverse problems of hyperspectral unmixing. The proposed approach models the main spectral variability (i.e., scaling factors) generated by variations in illumination or typography separately by means of the endmember dictionary. It then models other spectral variabilities caused by environmental conditions (e.g., local temperature and humidity and atmospheric effects) and instrumental configurations (e.g., sensor noise), and material nonlinear mixing effects, by introducing a spectral variability dictionary. To effectively run the data-driven learning strategy, we also propose a reasonable prior knowledge for the spectral variability dictionary, whose atoms are assumed to be low-coherent with spectral signatures of endmembers, which leads to a well-known low-coherence dictionary learning problem. Thus, a dictionary learning technique is embedded in the framework of spectral unmixing so that the algorithm can learn the spectral variability dictionary and estimate the abundance maps simultaneously. Extensive experiments on synthetic and real datasets are performed to demonstrate the superiority and effectiveness of the proposed method in comparison with the previous state-of-the-art methods.
Subfields: Ecology, Media Technology, Remote Sensing, Signal Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Ecology
- Jocelyn Chanussot
- Qian Du
- Swalpa Kumar Roy
- Danfeng Hong
- Jing Yao
- Bing Zhang
- Pedram Ghamisi
- Lorenzo Bruzzone
In Media Technology
- Jocelyn Chanussot
- Danfeng Hong
- Antonio J. Plaza
- Pedram Ghamisi
- Swalpa Kumar Roy
- Qian Du
- Jon Atli Benediktsson
- Jing Yao
In Remote Sensing
In Signal Processing
- Antonio J. Plaza
- Qian Du
- Lorenzo Bruzzone
- Jocelyn Chanussot
- Renlong Hang
- Pedram Ghamisi
- Danfeng Hong
- Behnood Rasti
Citations in the corpus
- 3D Instance Segmentation of MVS Buildings (2021, methodology): “D models may suffer from various degradation (such as noises), so it is worth investigating the robustness of our method in such degraded scenarios [69].”
- Deep learning for change detection in remote sensing: a review (2022, background): “What’s more, there una-voidably exists a problem of spectral variability in more hyperspectral imagery (Hong et al. 2018), which leads to inaccurate change detection results.”
- Deep Learning for Unmanned Aerial Vehicle-Based Object Detection and Tracking: A survey (2021, background): “Rapid movement changes in the external environment (such as light, cloud, fog, and rain, among others) cause aerial images to be fuzzy and noisy, which inevitably leads to image degradation [32].”
- Classification of Hyperspectral and LiDAR Data Using Coupled CNNs (2020, background): “In the past few years, hyperspectral data have been widely explored for this task [1]–[3].”
- Attention-Based Adaptive Spectral–Spatial Kernel ResNet for Hyperspectral Image Classification (2020, background): “The use of spectral information alone is difficult to recognize the structure types of surface materials in urban analysis [25].”
- Multimodal Fusion Transformer for Remote Sensing Image Classification (2022, background): “8, 9, 10, 11, 12, and 13 depict the obtained classification maps for the hyperspectral datasets i.e., Houston, Trento, MUUFL and Augsburg along with their corresponding other modalities like LiDAR, MS image, SAR and DSM data, respectively.”
- Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects (2021, background): “Considering the aforementioned limitations, HSI analysis is categorized into the following main streams: dimensionality reduction [6]–[10], spectral unmixing [11]–[17], object/change detection [18]–[22] classification [23]–[26], feature learning for”
- More Diverse Means Better: Multimodal Deep Learning Meets Remote-Sensing Imagery Classification (2020, background): “For instance, in urban planning, the structure types of surface materials are hardly identified using only one modality information (e.g., spectral data) [13].”
- Feature Extraction for Hyperspectral Imagery: The Evolution From Shallow to Deep: Overview and Toolbox (2020, neutral): “, sparse and collaborative representations [69]–[71].”