Quantification and characterizing of soil microstructure features by image processing technique

Chao Sheng Tang, Luan Lin, Qing Cheng, Cheng Zhu, Dong Wei Wang, Zhu Yuan Lin, Bin Shi

Research output: Contribution to journalArticlepeer-review

39 Scopus citations


An analyzing program SMAS based on digital image processing technique is developed for quantifying soil microstructure. By using SMAS, a series of geometrical and morphological indexes of soil particles/pores in microscale can be quantitatively determined. Three examples of using SMAS to quantify the microstructure features are shown. The analyzing results indicate that the developed program can effectively identify the morphology of soil particle and pore and accurately extract the soil microstructure indexes. A classification criterion for particle shape category is proposed based on the obtained values of morphology ratio and roundness. Moreover, effects of magnification and observation area of SEM images on the quantitative analysis results are discussed. It is important to select an appropriate magnification and observation area can cover as much structural information as possible while with high imaging quality. A recommendation approach is to stitch several images with relatively high magnification to one large image for quantification. Moreover, performing multiple scans on different zones of interest and then making comparative analysis is also an effective way to reduce quantification errors in microstructure observation. The findings of this investigation would be valuable for improving the reliability of quantitative characterization of soil microstructure on the basis of SEM images.

Original languageEnglish (US)
Article number103817
JournalComputers and Geotechnics
StatePublished - Dec 2020
Externally publishedYes

All Science Journal Classification (ASJC) codes

  • Geotechnical Engineering and Engineering Geology
  • Computer Science Applications


Dive into the research topics of 'Quantification and characterizing of soil microstructure features by image processing technique'. Together they form a unique fingerprint.

Cite this