Wavelength-aware 2D Convolutions for Hyperspectral Imaging

TL;DR

This paper introduces a wavelength-aware 2D convolution method tailored for hyperspectral imaging, addressing challenges like small datasets, large channel dimensions, and camera variability, while enabling data-driven camera filter learning.

Contribution

The paper presents a novel 2D convolution approach optimized for hyperspectral data, incorporating model bias and continuous channel definition, with the ability to learn camera filters in a data-driven manner.

Findings

Outperforms existing models in hyperspectral classification tasks

Enables learning of camera-specific filters from data

Facilitates design of optimal hyperspectral cameras

Abstract

Deep Learning could drastically boost the classification accuracy for Hyperspectral Imaging (HSI). Still, the training on the mostly small hyperspectral data sets is not trivial. Two key challenges are the large channel dimension of the recordings and the incompatibility between cameras of different manufacturers. By introducing a suitable model bias and continuously defining the channel dimension, we propose a 2D convolution optimized for these challenges of Hyperspectral Imaging. We evaluate the method based on two different hyperspectral applications (inline inspection and remote sensing). Besides the shown superiority of the model, the modification adds additional explanatory power. In addition, the model learns the necessary camera filters in a data-driven manner. Based on these camera filters, an optimal camera can be designed.