Demosaicking system and method for color array based multi-spectral sensors

摘要

A computer-implemented demosaicking system and method that can receive an image (or many images that represent individual frames of a video) at a demosaicking processor from a multi-spectral band camera. The image can include four or more band images that each correspond to an unique spectral band obtained by the multi-spectral band camera. A clustering module can perform spectral clustering of the four or more band images to identify multiple clusters. For each of the plurality of clusters, a weights module can determine a cluster weight by computing correlations between each of the unique spectral bands in each cluster. A super-resolution module can perform super-resolution for each of the unique spectral bands by utilizing the cluster weights from the weights module. The super-resolution module can iteratively apply the super-resolution for each of the unique spectral bands and a value for each unique spectral band can be updated after each iteration.

主权项

1. A computer-implemented demosaicking method, comprising the steps of:
receiving an image at a demosaicking processor, wherein the image comprises four or more band images that each correspond to an unique spectral band obtained by a multi-spectral band camera; performing spectral clustering of the four or more band images to identify a plurality of clusters with a clustering module, wherein the clustering module is implemented in a computer system that comprises instructions stored in a machine-readable medium and a processor that executes the instructions; for each of the plurality of clusters, determining a cluster weight by computing correlations between each of the unique spectral bands in each cluster with a weights module, wherein the weights module is implemented in a computer system that comprises instructions stored in a machine-readable medium and a processor that executes the instructions; and performing super-resolution for each of the unique spectral bands utilizing the cluster weights from the weights module with a super-resolution module, wherein super-resolution is applied iteratively for each of the unique spectral bands and a value for each unique spectral band is updated after each iteration, and wherein the super-resolution module is implemented in a computer system that comprises instructions stored in a machine-readable medium and a processor that executes the instructions.