On the Acquisition of High-Quality Digital Images and Extraction of Effective Color Information for Soil Water Content Testing

Abstract

Soil water content (SWC) is a critical indicator for engineering construction, crop production, and the hydrologic cycle. The rapid and accurate assessment of SWC is of great importance. At present, digital images are becoming increasingly popular in environmental monitoring and soil property analysis owing to the advantages of non-destructiveness, cheapness, and high-efficiency. However, the capture of high-quality digital image and effective color information acquisition is challenging. For this reason, a photographic platform with an integrated experimental structure configuration was designed to yield high-quality soil images. The detrimental parameters of the platform including type and intensity of the light source and the camera shooting angle were determined after systematic exploration. A new method based on Gaussian fitting gray histogram for extracting RGB image feature parameters was proposed and validated. The correlation between 21 characteristic parameters of five color spaces (RGB, HLS, CIEXYZ, CIELAB, and CIELUV) and SWC was investigated. The model for the relationship between characteristic parameters and SWC was constructed by using least squares regression (LSR), stepwise regression (STR), and partial least squares regression (PLSR). Findings showed that the camera platform equipped with 45° illumination D65 light source, 90° shooting angle, 1900~2500 lx surface illumination, and operating at ambient temperature difference of 5 °C could produce highly reproducible and stable soil color information. The effects of image scale had a great influence on color feature extraction. The entire area of soil image, i.e., 3,000,000 pixels, was chosen in conjunction with a new method for obtaining color features, which is beneficial to eliminate the interference of uneven lightness and micro-topography of soil samples. For the five color spaces and related 21 characteristic parameters, RGB and CIEXYZ spaces and characteristic parameter of lightness both exhibited the strongest correlation with SWC. The PLSR model based on soil specimen images ID had an excellent predictive accuracy and the best stability (R2 = 0.999, RMSE = 0.236). This study showed the potential of the application of color information of digital images to predict SWC in agriculture and geotechnical engineering.

Shadow removal method of soil surface image based on GAN used for estimation of farmland soil moisture content

“…Table 5 shows the overall performance of these methods. Comparing the visual results, the shadow image removal effects of these seven models in simple scenes such as image (1), image (2) and image (3) are relatively stable, and there are great differences in the performance of shadow images removal which are full of details such as image (4) and image (5). The following describes the performance of soil shadow removal of various technologies in order.…”

“…In image (2), image (3) and image (4), results show that this method does not seem to work on soil shadow removal, image (2) only deepens the color of shadow. In image (1) and image (5), more obvious shadow boundaries and shadow areas can be seen, the shadow removal effect is poor.…”

“…However, for the performance of image (4), although these three methods are in addition to the shadow of crops to a certain extent, the shadow generated by large soil aggregates on the right side of the image has not been effectively removed, especially CA NET seems can not to deal with such details of the shadow. Meanwhile, in image (5), when CA NET processes the clod shadow on the left side of the image and the semicircle shadow on the lower right corner, the color of the shadow area is changed, a red area appears. Moreover, the color of the output image of STC GAN and CA NET is darker than the original shadow image, so we prefer to get an output image which is close to the original color.…”

“…Rapid SMC detection is important for precision agriculture management. Computer vision and soil images can be used to estimate soil moisture [2][3][4][5]. Under the same soil texture and fertility, the Gray value of soil surface image [6], soil apparent electrical conductivity (ECa) [7], color parameters [8,9], and soil mechanical resistance (MR) [10] can be used for data fusion to rapid estimate surface soil moisture.…”

Optimizing Color-Difference Formulas for 3D-Printed Objects

“…Visual differences and applicability of current color-difference formulas for 3D sample pairs must be more carefully studied, as is currently being done by CIE Technical Committee 8-17, "Methods for Evaluating Color Difference between 3D Color Objects" [17]. Hopefully, Sensors 2022, 22, 8869 2 of 15 a successful prediction of visual color differences for 3D objects will be useful for industry and in different applications [18][19][20][21].…”