The article “A General Method for Estimating Fluorescent Donaldson Matrices” by Shoji Tominaga, Keita Hirai, and Takahiko Horiuchi, published in 2018, presents a generalized method for estimating the bispectral Donaldson matrices of fluorescent objects. Fluorescence in materials has gained prominence in everyday life due to its ability to enhance the visual appearance of surfaces, making them appear brighter and more vivid compared to non-fluorescent reflective materials. This quality is quantitatively described by the bispectral radiance factor, a function that depends on the excitation wavelength of the incident light and the emission/reflection wavelength.
To overcome the limitations of traditional two-monochromator methods, which are costly and confined to laboratory environments, Tominaga et al. had previously developed a method using two illuminant projections. However, these two-illuminant-based systems faced reliability and stability issues under low-intensity sources or noisy conditions. An improvement was required to make the process more viable and accurate outside the laboratory.
The main innovation of the 2018 article is a generalized method that employs two or more light sources and segments the visible wavelength range into components of reflection, fluorescent excitation, and fluorescent emission. It observes the matte surface of a fluorescent object, uniformly illuminated, through a spectral imaging system with narrowband characteristics in the visible range. The estimation problem is formulated as an optimization problem to minimize the residual error of the captured observations.
The presented algorithm iteratively estimates three spectral components: fluorescent excitation, fluorescent emission, and reflection. The procedure divides the visible wavelength range into two parts: one that includes only reflection and another that also contains fluorescent emission. In the first case, the reflection component is estimated directly, while in the second case, the three components are iteratively estimated by minimizing residual errors under different illuminants.
The proposed method is robust, given that the estimates minimize the average residual error of the observations. Its viability is examined through experiments with various fluorescent samples and different illuminants, showing detailed results for pink and green fluorescent paper. The spectral characteristics of the reflected forms, spectral emission, and fluorescent emission range are calculated and compared with conventional methods. The optimization focuses on minimizing the performance index J, which reflects the average residual error. Experimental results illustrate that the proposed method offers more accurate estimates and lower error indices compared to previous methods, constituting a significant advance for the estimation of Donaldson matrices in practical applications.
In conclusion, this article proposes and validates a generalized and effective method to accurately estimate bispectral Donaldson matrices of fluorescent objects, using multiple light sources and an optimized approach that overcomes the restrictions and limitations of traditional methods.
