Using Digital Image Analysis to Compare Soil Moisture Estimation Models.

Many researches present a plethora of methods for measuring the soil moisture content; yet, the least time consuming method is the image analysis. Though the estimation of soil moisture content is rapid, the established technology is not fully developed; each moisture content estimation model shows both advantages and weaknesses.

For this study, sand soil samples were used in five depths and were placed into water to saturate the soil moisture content. After the samples reached the saturation state, they dried naturally in the dark room. And using HYPROP to measure the soil moisture content, which is a water retention curve analyzer. Moreover, its status from the saturated to the dry phase was also recorded. After the digital images of the soil samples were taken by digital camera in the dark room, the images were segmented by the Photoscape software. The software ImageJ was used to analyze the hue of the images – R (red), G (green), and B (blue). Furthermore, the correlation between the soil moisture content and the digital image hue was also investigated.

The correlation between the soil moisture content and the digital image hue was used to compare the GBHS, RGB, and SV models (Liu Weidong et al. (2005); Persson (2005)). By investigating the significant correlation and by receiving more accurate estimations, the soil moisture content measurement could become more effective.

The results showed that these three models have comparably good results; whereas, the SV model was the best for the estimations based on the average values of Hue (H), Saturation (S), and Value (V); therefore, this method is feasible and could estimate a large range of soil moisture in short time. Additionally, it could notably reduce the required resources for such measurements in the future.

Keywords: Digital images, Hue analysis, Soil moisture content