Spectral optimization of illumination for contrast enhancement in machine vision

This study explores the potential of algorithmically optimized illuminants to improve contrast between objects for machine vision. We develop multiple algorithms to generate these spectra and introduce a light source capable of producing the optimized illuminants. Recognizing that the optimal illuminant depends on the observer, we also propose a straightforward method to estimate a camera's spectral sensitivity using the designed programmable light source. Through systematic evaluation, we demonstrate that the effectiveness of the algorithms varies depending on the types of color pairs being analyzed. Finally, we present a case study to illustrate the real-world applicability of our approach for contrast enhancement tasks.