Iterative spectral correlation based multispectral image demosaicking

Multispectral imaging systems with a multispectral filter array (MSFA) provide an affordable and portable way to capture multispectral images (MSIs) that have a variety of applications in different fields. These systems generate raw images initially and require an effective multispectral image demosaicking technique for reconstructing the MSI from the raw data. These demosaicking methods require proper usage of the spectral correlation available between the bands of MSI to generate high quality MSI. However, the existing demosaicking methods only partially utilize this spectral correlation, as they use spectral correlation only in the initial estimation of MSI. In this work, we utilize the spectral correlation between bands iteratively and finally enhance the quality of the generated image using median filtering based image enhancement. The exploratory results on the two standard datasets publicise the quality of the presented method on the various metrics.