Method for the numerical evaluation of the color rendering quality of matrix photodetectors

Abstract

To objectively evaluate the color image of an architectural object when changing its spectral quality and luminance in lighting, it is necessary to ensure its accurate fixation. The measurement of the color coordinates of an object provides the fully capture of its appearance, but this does not allow to get an adequate evaluation of the visual impression without visualizing the lighting scene. This requires a thorough procedure of reproducing the same color coordinates for each point of the image and the object with the same angular coordinates. In this work, an attempt is made to develop a numerical criterion for evaluating the quality of color rendering of the camera as a fixer of color visual impression of the selected scene. The development of a numerical detailed method for evaluation of the quality of color rendering of cameras will allow capturing more reliably such a subjective concept as the correspondence of the visual impression of the real scene and the image file of this scene. The digital format, which contains information about the object, avoids the problems associated with aging of the image. The proposed method of developing ways to numerically evaluate color distortion in photography is considered on the example of digital cameras Nikon D300s, Sony DSC-H5. The described approach for the case of known spectral characteristics allows to unambiguously link the calculated reaction of the camera with the chromaticity coordinates of spectrally pure colors. Modern methods of evaluating the quality of light sources (IES TM-30-15) allow evaluation of the direction of the shift. The proposed indicator – graph E (λi) numerically characterizes the difference between the reaction of the camera and the reaction of the human eye and does not give an idea in which direction there is a difference, conditionally red or blue image compared to the natural scene, but is the simplest and most understandable to the untrained user. It is shown that with an accuracy of measuring spectral characteristics of 1% for the considered example with Nikon D300s, this indicator at different wavelengths is calculated with an uncertainty of not more than 0.002.