Novel Spectral Loss Function for Unsupervised Hyperspectral Image Segmentation

Neural networks (NNs) have gained importance in hyperspectral image (HSI) segmentation for earth observation (EO) due to their unparalleled data-driven feature extraction capability. However, in many real-life situations, ground truth is not available, and the performance of unsupervised NNs is still susceptible to enhancement. To overcome this challenge, this letter presents a new loss function to improve the performance of unsupervised HSI segmentation models. The spectral loss function, Sl, which can be included in different models, is based on the purity of the unmixing endmembers and the spectral similarity of the clusters provided by the NN to determine the classes. It is incorporated into a 3-D convolutional autoencoder (AE) to validate its performance on four standard HSI benchmarks. Furthermore, its performance has been qualitatively examined in a real case study, an oil spill without ground truth. The results show that Sl is a breakthrough in unsupervised HS segmentation, obtaining the best overall performance and highlighting the importance of spectral signatures. Additionally, the dimensional reduction is also vital in compacting the spectral information, which facilitates its segmentation. The source code is available at https://github.com/mhaut/HSI-3DSpLoss.

Automated summary

This letter presents a new loss function to improve the performance of unsupervised hyperspectral image (HSI) segmentation models. The spectral loss function, Sl, based on the purity of the unmixing endmembers and the spectral similarity of the clusters, is incorporated into a 3-D convolutional autoencoder to validate its performance. The results show that Sl is a breakthrough in unsupervised HSI segmentation, highlighting the importance of spectral signatures.