Accession Number : ADA334849
Title : Characterization of Particle Assemblies Through Digital Image Processing
Descriptive Note : Final rept. 1 Oct 96-30 Sep 97
Corporate Author : MICHIGAN UNIV ANN ARBOR
Personal Author(s) : Hryciw, Roman D. ; Ghalib, Ali
PDF Url : ADA334849
Report Date : 24 DEC 1997
Pagination or Media Count : 16
Abstract : Imaging techniques are proving to be a viable alternative to mechanical sieving for determination of soil grain size distribution. While such distributions are relatively easily obtained when the soil grains are non-contacting, interpretation of in-situ images of contacting grains (assemblies) is considerably more difficult. As such, two approaches were developed for interpreting such images including: edge detection and completion by Hough transforms with active contouring; and pixel density analysis. A reasonably precise measure of grain size can be obtained using circular Hough transforms in conjunction with active contouring. However, the method is computationally very intensive and has only been tested on highly idealized assemblies of soil grains. Edge pixel densities (EPD) and their coefficients of variation with increasing sampling window size provide a rapid means for assessing grain sire and uniformity and for detection of soil interfaces within an image.
Descriptors : *IMAGE PROCESSING, *SOIL COMPACTION, GRAIN SIZE, PATTERN RECOGNITION, COMPUTER VISION, PIXELS, INTEGRAL TRANSFORMS.
Subject Categories : Cybernetics
Distribution Statement : APPROVED FOR PUBLIC RELEASE