Optimization of a Multispectral Imaging System

Abstract

We have previously presented work using a simple computational model of a multispectral imaging system which we have used to investigate the effect of various parameters (such as the number and properties of the colour channels) on the accuracy of spectral estimation. It was shown that the parameters of a multispectral imaging system interact with each other in a complex way and therefore the optimum set of parameters cannot easily be determined. This paper concerns the development of an optimization technique (based upon simulated annealing) to allow the parameter space of the imaging system to be searched to find the system parameters that give the lowest reconstruction error. Initial results from the optimization analysis show that the optimum spectral properties of the colour channels depend both upon the illuminant under which the image is captured and the goal of the spectral reconstruction process.