Enhancement of spectral matching imager output for discriminating multiple similar spectra by time-nonlinear wavelength sweeping

A method for enhancing the difference in the outputs of the spectral matching imager (SMI) for multiple similar spectra is proposed. A spectral weight function is introduced to the joint probability density of the target spectra for diagonalizing the covariance matrix by sweeping the peak wavelength in the SMI nonlinearly to time. The weight function, or equivalently a wavelength sweep function, is optimally obtained by minimizing a cost functional consisting of the sum of all the squared off-diagonal elements of the covariance matrix. The experimental results on a set of three similar spectral reflectances demonstrate that the proposed method effectively suppresses the off-diagonal elements of the covariance matrix with the estimated weight function to enhance the difference in SMI outputs.