A generic multispectral demosaicking method based on inter-channel spectrum correlation

Multispectral images have more spectral bands compared to RGB color images and also contain more information. Therefore, multispectral images have a wide range of applications, including atmospheric environment detection, medical diagnostics, agricultural disease detection, and food monitoring. Because acquiring raw multispectral images requires embedding multiple cameras and various mechanical and optical components to image them once, these systems are very heavy and not inexpensive, so their use is limited. In order to solve these problems, a kind of sensor technology called Multispectral Filter Array (MSFA) has become a hot topic and gradually developed. Snapshot multispectral imaging based on MSFA can lead to severe sparsity in each channel of the original captured image. Each pixel on the detector based on MSFA only responds to the optical information of the corresponding spectral band, so the other spectral band information of the pixel is lost. We refer to the original image captured based on MSFA as mosaic image. In order to obtain multiple single spectral band images with full resolution, we need to carry out a process called demosaicking on the original image. In this article, we present a generic demosaicking approach based on inter-channel spectrum correlation. In order to verify the effect of the proposed approach, we conducted some validation experiments on a public database. Experimental results show that the proposed approach is better than the classical methods in spatial reconstruction and spectral accuracy.