A Snapshot Imaging System and Reconstruction Algorithm for Hyperspectral Light Field

Abstract

The invention discloses a snapshot type imaging system and a reconstruction algorithm of a hyperspectral light field. The snapshot type imaging system comprises a beam splitter, a controller, a first observation channel and a second observation channel. The beam splitter is used for splitting incident light of a scene into two paths, and the two paths of light enter the first observation channel and the second observation channel respectively. The first observation channel and the second observation channel are used for acquiring observation results of the scene respectively. The controller is used for reconstructing a five-dimensional hyperspectral light field according to the acquired observation results and the correlation between angle dimensionality and spectrum dimensionality in hyperspectral light field signals. According to the technical scheme, the five-dimensional hyperspectral light field is optimized and reconstructed through the correlation between the angle dimensionality and the spectrum dimensionality in the light signals, and a good reconstruction effect can be achieved from seriously undersampled observation.

Description

technical field [0001] The invention relates to the technical field of computational imaging, more specifically, to a hyperspectral light field snapshot imaging system and a reconstruction algorithm. Background technique [0002] In recent years, computational imaging has developed rapidly due to the advancement of optical instruments and the growth of computing power. The ultimate goal of computational imaging is to simultaneously acquire seven-dimensional plenoptic information, including three-dimensional space (two-dimensional plane and one-dimensional depth), one-dimensional time, one-dimensional spectrum and two-dimensional angle. Traditional imaging methods focus on two-dimensional planar space (ie, traditional image) and one-dimensional time (ie, traditional video), which provide limited information. Current research on other dimensions includes depth / 3D imaging, multispectral / hyperspectral imaging, and light field imaging. At the same time, portable depth cameras a...

AI Technical Summary

Problems solved by technology

Such scanning methods take advantage of the spatial correlation of images to a certain extent and achieve higher spatial and spectral resolutions, but they are not suitable for scenes that require real-time shooting, such as scenes containing dynamic objects or lighting changes

Images