Adaptive band selection snapshot multispectral imaging in the VIS/NIR domain

TL;DR

This paper proposes an adaptive hyperspectral band selection method for snapshot multispectral imaging in the VIS/NIR domain, optimizing filter parameters to enhance target detection in real-time applications.

Contribution

It introduces a novel adaptive band selection approach that maximizes detection contrast by optimizing filter positions and linewidths for snapshot imagers.

Findings

Simulations show improved detection contrast with adaptive band selection.

The method enables real-time target detection using snapshot multispectral imaging.

Adaptive filters can compete with traditional hyperspectral imagers in detection efficiency.

Abstract

Hyperspectral imaging has proven its efficiency for target detection applications but the acquisition mode and the data rate are major issues when dealing with real-time detection applications. It can be useful to use snapshot spectral imagers able to acquire all the spectral channels simultaneously on a single image sensor. Such snapshot spectral imagers suffer from the lack of spectral resolution. It is then mandatory to carefully select the spectral content of the acquired image with respect to the proposed application. We present a novel approach of hyperspectral band selection for target detection which maximizes the contrast between the background and the target by proper optimization of positions and linewidths of a limited number of filters. Based on a set of tunable band-pass filters such as Fabry-Perot filters, the device should be able to adapt itself to the current scene and the target looked for. Simulations based on real hyperspectral images show that such snapshot imagers could compete well against hyperspectral imagers in terms of detection efficiency while allowing snapshot acquisition, and real-time detection.