Color calibration of moist soil images captured under irregular lighting conditions

Abstract

Soil color is a valuable indicator that visually reflects soil properties. In the field of soil science, soil color has been widely used for soil identification and property assessment. Soil color is influenced not only by soil properties but also by lighting conditions; therefore, adequately considering the effect of lighting conditions for the color-based soil assessment is essential. This paper proposes a color calibration method for moist soil images captured under irregular lighting conditions. Four samples of weathered granite soil with varying water contents were captured under nine different lighting conditions. The effect of lighting conditions on the color of moist soil was analyzed using linear regression analysis in the CIELAB color space. The findings demonstrate a linear relationship between soil color and lighting conditions. Furthermore, the slopes of the linear regression equations remained consistent across all moist soil samples, irrespective of soil type and water content. Based on this, calibration equations were derived to adjust the color of moist soil images captured under arbitrary lighting conditions to those under the desired lighting conditions. The proposed method was validated using soil images captured under natural lighting conditions, which show high accuracy in color calibration except for cases with illuminance exceeding 70,000 lux. This color calibration method provides a reliable approach to evaluating soil properties through accurate color representation and presents new possibilities for developing digital image-based soil property prediction techniques that are less sensitive to lighting conditions.