Abstract:

Silty clay is a typical soft soil and characterized with its coagulation structure. In the field, it is very difficult to directly determine the shape and size of a grain in the silty clay by eyes. Some colorful photographs of the silty clay were captured by a commercial camera and transformed into grayscale images. To these grayscale images, the six histogram-based parameters, including mean, standard deviation, smoothness, third order moment, energy and entropy, were thereafter obtained. Considering the spatial relations of grayscale intensity in an image, five parameters, including angularity second order moment, contrast, correlation, inverse difference moment and entropy, are also obtained based on a gray co-occurrence matrix. Furthermore, the regression technique was used to estimate the relations between the texture parameters and the traditional engineering properties of silty clays. It is shown that these relations are much satisfactory and may be used in quickly examining the engineering properties of silty clay. The method developed is referable to the geotechnical design in soft grounds.

Key words: silty clay, texture, image analysis, histogram, gray co-occurrence matrix, multiple linear regression