Estimation of pure water transparency using a calculated image

Abstract

The transparency of pure water is the maximum of all transparencies; it is a value required for transparency analysis. However, it is difficult to determine the pure water transparency value directly through experiments, and no universal value has been obtained theoretically thus far. Therefore, the transparency of a colour image is estimated by observing it on a luminance-calibrated LCD screen that displays the tristimulus values calculated for light arriving from the direction of a Secchi disk and background. The light scattered by water molecules was approximated to the first order. For incident angles of 0°, 10°, 20°, ..., 80°, and 89°, the colour-image transparencies (CITs) were in the range of 101 ± 2.6 m to 124 ± 1.1 m. The monochromatic image transparencies (MITs) observed were similar to the research and the transparency-judgment mechanism was in the range of 95 ± 1.3 m to 117 ± 3.4 m. The total error was determined as the sum of the 95% confidence interval and trueness with respect to the transparency of light, calculated using the first-order scattering approximation. The root mean square of the relative error to CITs of the automatically determined transparency based on colour difference and threshold was 6.3 % while that to MITs based on contrast and threshold was 3.6 %. To improve the accuracy of colour-imagetransparency measurements, it is necessary to increase the order of calculating molecular scattering, improve image observation accuracy, and establish an evaluation standard. One methodology considered is to obtain an approximate pure water transparency value using a scaled-down transparency measurement experiment based on adding an absorbent to pure water.