The Relationship Between The Physical Properties Of Soil, Fractal Dimension, And Shape Factors Of Its Fragmented Aggregates: A Two-dimensional Digital Image Processing And Analysis Approach

Soil physical properties are used for soil classification and hence serve as the first estimator of almost every mechanical and hydrologic soil behavior, but the usual laboratory testing method of estimating these physical properties is time consuming, burdensome, and somehow subjective, hence posing a challenge to geotechnical engineers and soil lab technicians.To minimize these shortfalls to an acceptable level and make the determination of soil physical properties more objective, this study investigates the possibility of using digital image processing and analysis(DIP&A) approach to estimate soil physical properties in place of the most widely used laboratory approach by establishing relationships between the physical properties of soil(water content, density-wet bulk and dry, porosity, and void ratio) and the image features of its fragmented aggregates identified as fractal dimension(FD) and shape factors recognized as circularity, aspect ratio, and roundness in the study. The aggregates shape of soil are complex and irregular. Fractal theory uses the concept of FD as a way to describe the shape of irregular particles or objects and shape factors are used to describe the overall geometric characteristics of an object or particle.In this study, samples of soil were collected and laboratory test performed as per the American Society of Testing and Materials(ASTM) Standards to determine the physical properties and the same samples fragmented into appropriate sizes and their images acquired for image processing and analysis to determine the image features. The physical properties of soil were statistically correlated and regressed against its image features using appropriate regression model with the correlation coefficient(r) as the objective measure and premise of the correlation. The results of the digital image analysis were verified by conducting laboratory test and image analysis of soil sample collected from different site other than the one previously used to develop the relation models and the new values of the validated sample replaced in the developed models in order to find a percentage of error between conventional laboratory testing and the proposed new method DIP&A.The outputs of the regression analysis were curvilinear models that represent the relationship between soil physical properties and its image features as identified in the study. The r value varies from 0.3 to 0.4 for the relationship between physical properties and FD, and for the relationship between physical properties and shape factors, the r value ranges from 0.5 to 0.6. The new values of the FD and the average shape factors of the validated sample were replaced in the developed relation models resulting in a reasonable percentage of error between conventional laboratory testing and DIP&A method that ranges from 0.60% to 3.57% for the relationship between physical properties and FD, and an average percent error that varies from 1.75% to 5.13% for the relationship between physical properties and shape factors.The results indicate that DIP&A method proposed in the study appears to be a useful and promising method for estimating or determining the physical properties of soil and was also shown to be a valuable tool for quantifying the geometric properties of soil aggregates.