In this study, we developed a method that allows the delimitation of horizons and the quantification of coarse fragments in soil profiles from volcanic areas using digital images. The delineation of the soil profile horizons included the following phases: pre-processing of the digital image, extraction of color systems from the pre-processed image, k-means segmentation of the HSV and CIE L*a*b* color systems, delineation of the horizons, and determination of the horizon characteristics. The coarse fragments were quantified in the following three phases: superpixel analysis of the soil profile image, histogram classification of the image objects, and extraction and quantification of the coarse fragments. For horizon delineation, the HSV color system performed better for the Eutric Andic Cambisol (Loamic, Ochric), and the CIE L*a*b* system performed better for the Eutric Skeletic Mollic Silandic Andosol (Loamic). The RGB image and the S component of the HSV system demonstrated similar coarse fragment volume calculation performance for the Eutric Andic Cambisol (Loamic, Ochric), whereas the S component worked best for the Eutric Skeletic Mollic Silandic Andosol (Loamic). We created a graphic decision-making system for the delineation of soil horizons and for the quantification of coarse fragments in digital images of soil profiles.

在这项研究中，我们开发了一种方法，该方法可以使用数字图像对火山区土壤剖面中的视界进行定界和量化。土壤剖面视野的划分包括以下几个阶段：数字图像的预处理，从预处理图像中提取颜色系统，HSV和CIE L * a * b *颜色系统的k均值分割，描绘层的确定以及层特征的确定。在以下三个阶段对粗碎片进行定量：土壤剖面图像的超像素分析，图像对象的直方图分类以及粗碎片的提取和定量。对于地平线描绘，HSV彩色系统对Eutric Andic Cambisol（Loamic，Ochric）的效果更好，CIE L * a * b *系统对Eutric Skeletic Mollic Silandic Andosol（Loamic）的效果更好。HSV系统的RGB图像和S分量对Eutric Andic Cambisol（Loamic，Ochric）表现出相似的粗片段体积计算性能，而S分量对于Eutric Skeletic Mollic Silandic Andosol（Loamic）则表现最佳。我们创建了一个图形决策系统，用于划定土壤层位并量化土壤剖面数字图像中的粗碎屑。