This research project to determine the best geometry and best processing algorithms for implementation of multispectral focal-plane array cameras, based on a mosaicked technology, for use in missiles and smart ordnance. In a mosaicked sensor, each pixel detector is covered with a wavelength-specific optical filter. Since only one spectral band is sensed at each pixel, the other bands must be estimated from neighboring pixels. In the commercial color digital camera market, sophisticated algorithms have been developed to perform this estimation algorithms which are not appropriate for more than three spectral bands and not optimized for the Infrared (IR). Currently, mosaicked sensors in the IR do not exist, although the technology to produce them is available. The results from this project will serve as a guide to design and production of such devices.
This proposal shows that mosaicked arrays are the most effective option for military applications such as missiles and smart ordnance; however, mosaicked arrays require data interpolation. We propose to develop optimal algorithms for interpolating the unmeasured spectral bands as well as algorithms for interpolating missing multispectral data in the case that detectors are defective. Detector geometry is also a concern, and we propose to investigate several alternative detector arrangements for three to seven bands, and to determine the geometry which optimizes signal quality and accuracy of representation. Effective display of multispectral IR information is an essential component of any system utilizing such data. As part of this project, we will develop optimal techniques for displaying such data on conventional color displays.
The project is administered by the College of Engineering at North Carolina State University with a subcontract to the University of Tennessee, Knoxville.