Global Journal of Human-Social Science, B: Geography, Environmental Science and Disaster Management, Volume 22 Issue 3

d) Image Comparison As a complementary analysis and a way of using data science tools for image analysis, images were separated from the fine-earth fraction of soils (in this case the material passing in sieve no. 10 and retained in no. 200) made in the Stereo Microscope Zeiss Discovery V8 in order to compare the soils to each other and to verify some similarity with the results obtained in the clustering. The mineralogy of the fraction passing in sieve no. 200 was studied by performing the X-ray Diffraction test and images of the coarse fraction of the soil (retained in sieve no. 10) were also extracted, but were not considered in the analyses of this article. In summary, the process of comparing images in Python is done by implementing a function that receives the image and processes it by transforming it into vectors. The function applied in this article was the BIC (Border-Interior Pixel Classification), which transforms and rescales the image, and finally computes two color histograms: one for the "interior" pixels and the other for the "border" pixels (edges of the image), then normalizes the histograms and concatenates them into a single vector. Vectors represent the frequencies of colors in the image, which allows, with the aid of a distance measurement, to compare images for color similarity. To make the comparison, you must set an image as a reference so that the distance of the other images related to it is calculated. In this part of the analysis, we chose to use the same distance measurement used in the cluster analysis that was the Euclidean distance. III. R esults and D iscussion a) Geotechnical Parameters Table 5 presents part of the database of the studied soils: geotechnical classifications (TRB, USCS and MCT), MCT classification, real density, Atterberg limits, granulometry, mechanical compaction parameters, CBR (and expansion), MR, DP and chemical data. Table 6 presents the continuity of the database with parameters associated with the characteristic curve of the soils. The units of measurement of each parameter were presented in Table 2. Regarding the USCS class, five samples were classified as ML (silt soils of low compressibility), four were classified as SC (clay sand), and two as SM (silty sand). Soil 03 was classified as SM-SC and soil 13 was classified as CL (clay soil of low compressibility). Table 5: Classification database, true density, consistency limits, granulometry, compaction, CBR, expansion, MR, DP and chemical data from soils studied in this research Soil 1 2 3 4 5 6 7 8 9 10 11 12 13 MCT LG' LG' NA' LG' LA' LA' NA' LA' LA' LA' LA' LG' LA' TRB A-6 A-7-6 A-4 A-6 A-6 A-6 A-6 A-4 A-4 A-7-6 A-7-5 A-6 A-6 USCS ML ML SM-SC SC SC SM ML ML SM SC ML SC CL d' 247 139 133 87 131 140 147 107 109 217 188 150 154 e' 0,94 0,62 1,19 0,93 0,62 0,62 1,16 0,84 0,91 0,87 0,84 0,81 0,81 c' 2,0 1,9 1,2 1,7 1,4 1,05 1,05 1,29 0,92 1,05 1,0 1,67 1,1 PMI 75 10 153 58 9 10 143 40 58 56 49 40 40 PMI wo 75 0 98 38 9 0 34 0 0 0 0 39 42 δ 2,63 2,67 2,65 2,63 2,64 2,67 2,69 2,70 2,68 2,68 2,68 2,67 2,68 LL1 32,2 33,7 22,0 26,2 30,0 28,5 32,2 34,5 27,4 27,5 33,9 27,8 28,7 PL1 22,8 25,9 18,0 20,3 20,4 20,4 24,4 26,5 19,3 19,8 30,4 18,7 19,8 PI1 9,4 7,8 4,0 5,9 9,6 8,1 7,8 8,0 8,1 7,7 3,5 9,1 8,9 LL2 35,6 41,3 24,1 33,4 37,9 39,9 39,9 39,5 34,7 40,5 46,6 34,8 35,9 PL2 25,0 27,9 17,4 21,7 23,1 26,5 28,5 32,4 24,9 25,4 30,1 21,4 24,2 PI2 10,6 13,4 6,7 11,7 14,8 13,4 11,4 7,1 9,8 15,1 16,5 13,4 11,7 %Clay 50,2 47,8 32,3 38,8 41,7 42,6 46,5 47,0 42,0 40,3 45,5 34,9 44,7 %Silt 14,2 19,0 5,6 6,6 4,8 5,2 13,6 19,3 5,7 8,1 6,8 4,4 8,4 %FS 25,3 18,7 55,9 36,2 31,5 31,1 27,3 21,8 35,3 29,9 25,9 33,6 25,5 %MS 10,0 11,2 6,1 17,3 18,2 20,6 12,3 11,5 16,7 20,2 20,9 25,7 20,5 %CS 0,3 1,2 0,1 0,8 2,1 0,6 0,3 0,3 0,3 0,8 0,9 0,9 0,9 %Grav 0,1 2,1 0,0 0,2 1,7 0,1 0,0 0,1 0,0 0,8 0,1 0,4 0,1 ρ dmáx 1,79 1,75 1,97 1,91 1,94 1,88 1,78 1,69 1,90 1,87 1,86 1,94 1,86 W o 17,4 18,4 10,6 13,1 13,7 13,8 16,5 19,0 12,7 12,7 15,6 11,2 13,2 © 2022 Global Journals Volume XXII Issue III Version I 15 ( ) Global Journal of Human Social Science - Year 2022 B Clustering of Fine-Grained Tropical Soils using Data Science Tools Applied to their Geotechnical Properties