FT-IR imaging employing a focal plane array (FPA) detector is often plagued by low signal-to-noise ratio (SNR) data. A mathematical transform that re-orders spectral data points into decreasing order of SNR is employed to reduce noise by retransforming the ordered data set using only a few relevant data points. This approach is shown to result in significant gains in terms of image fidelity by examining microscopically phase-separated composites termed polymer dispersed liquid crystals (PDLCs). The actual gains depend on the SNR characteristics of the original data. Noise is reduced by a factor greater than 5 if the noise in the initial data is sufficiently low. For a moderate absorbance level of 0.5 a.u., the achievable SNR by reducing noise is greater than 100 for a collection time of less than 4 min. The criteria for optimal application of a noise-reducing procedure employing the minimum noise fraction (MNF) transform are discussed and various variables in the process quantified. This noise reduction is shown to provide high-quality images for accurate morphological analysis. The coupling of mathematical transformation techniques with spectroscopic Fourier transform infrared (FT-IR) imaging is shown to result in high-fidelity images without increasing collection time or drastically modifying hardware.
ASJC Scopus subject areas