Multispectral light-field imaging method

Abstract

The invention discloses a multispectral light-field imaging method. The adopted hardware comprises a wideband filter array, an isomerism camera array, a control panel array and an information joint processing device, wherein the wideband filter array, the isomerism camera array, the control panel array and the information joint processing device are arranged in sequence in the light path direction. The imaging method comprises the steps of placing wideband filters with different wavelengths among each camera lens and sensor of the isomerism camera array so that each camera in the isomerism camera array can receive spectral information with a fixed waveband ; using the information joint processing device to conduct stereo matching based on a convolutional neural network on multipath information obtained by the camera array to obtain angular information of incident ray and obtain optical field information within a full field range; conducting camera calibration and view field aligning on distribution positions of cameras, and obtaining multi-waveband spectral information of which the number is three times the number of cameras under any camera perspective in the camera array through spectral demultiplexing. According to the multispectral light-field imaging method, multispectral optical field information of each pixel with the full field range can be obtained simultaneously, and thus dynamic and united obtaining of optical field and spectral information is achieved.

Description

technical field [0001] The invention relates to the field of computer photography, in particular to a multispectral light field imaging method. Background technique [0002] Computational photography is an emerging research direction that deeply intersects computer vision, digital signal processing, graphics and other disciplines. Improve traditional cameras from the perspective of imaging mechanism, organically combine hardware design with software computing capabilities, break through the limitations of classic imaging models and digital cameras, enhance or expand the data collection capabilities of traditional digital cameras, and capture real-world scene information in an all-round way . Carry out related research work on scene light field spectral information, and obtain images with high spatial, temporal, spectral and directional resolution at the same time, which is of great significance for 3D reconstruction, security reconnaissance, digital entertainment and other ...

AI Technical Summary

Problems solved by technology

[0004] The existing hyperspectral technology expands the traditional imaging technology and can obtain multi-dimensional spectral information. However, according to different technical requirements and acquisition conditions, most acquisition systems compensate for spectral resolution by sacrificing spatial or temporal resolution, and lose depth information

Images