Multispectral snapshot demosaicing via non-convex matrix completion
This addresses the problem of data loss in multispectral imaging for applications like remote sensing and ground-based monitoring, representing an incremental improvement over existing demosaicing methods.
The paper tackles the problem of reconstructing full multispectral data cubes from severely undersampled snapshot mosaic sensors by using non-convex matrix completion techniques initialized with traditional demosaicing algorithms. The result shows a peak signal-to-noise ratio improvement of 2 to 5 dB over current state-of-the-art methods in simulations with a 16-frequency mosaic sensor across various scenes.
Snapshot mosaic multispectral imagery acquires an undersampled data cube by acquiring a single spectral measurement per spatial pixel. Sensors which acquire $p$ frequencies, therefore, suffer from severe $1/p$ undersampling of the full data cube. We show that the missing entries can be accurately imputed using non-convex techniques from sparse approximation and matrix completion initialised with traditional demosaicing algorithms. In particular, we observe the peak signal-to-noise ratio can typically be improved by 2 to 5 dB over current state-of-the-art methods when simulating a $p=16$ mosaic sensor measuring both high and low altitude urban and rural scenes as well as ground-based scenes.