Optical Codification Design in Compressive Spectral Imaging: From Mathematical to Deep Learning Optimization

Abstract

The optical codification in compressive spectral imaging has been an area of continuous improvement. The study of the coded apertures as modulation devices and their contribution to the final spectral image quality have allowed proposing different methods of coded aperture optimization. Four different coded aperture design approaches developed through the years are reported to show the improvement process, followed by the comparison of the resulting designs in the reconstruction of spectral images. Furthermore, the latest research on deep learning-based methods has resulted in additional tools for the designs, opening more opportunities in this area of research.