Approximate Message Passing in Coded Aperture Snapshot Spectral Imaging

TL;DR

This paper introduces AMP-3D-Wiener, a novel hyperspectral image reconstruction algorithm using approximate message passing with an adaptive Wiener filter, outperforming existing methods in accuracy and simplicity without parameter tuning.

Contribution

The paper presents a new AMP-based hyperspectral imaging reconstruction algorithm that employs an adaptive Wiener filter, eliminating the need for parameter tuning and improving performance.

Findings

AMP-3D-Wiener outperforms GPSR and TwIST in accuracy.

It requires less runtime for comparable results.

The method does not need parameter tuning.

Abstract

We consider a compressive hyperspectral imaging reconstruction problem, where three-dimensional spatio-spectral information about a scene is sensed by a coded aperture snapshot spectral imager (CASSI). The approximate message passing (AMP) framework is utilized to reconstruct hyperspectral images from CASSI measurements, and an adaptive Wiener filter is employed as a three-dimensional image denoiser within AMP. We call our algorithm "AMP-3D-Wiener." The simulation results show that AMP-3D-Wiener outperforms existing widely-used algorithms such as gradient projection for sparse reconstruction (GPSR) and two-step iterative shrinkage/thresholding (TwIST) given the same amount of runtime. Moreover, in contrast to GPSR and TwIST, AMP-3D-Wiener need not tune any parameters, which simplifies the reconstruction process.