Published Online: 17 May 2013
Page Count: 14
Research Assistant, Univ. of Michigan, Dept. of Civil and Environmental Engineering, Ann Arbor, MI
Hryciw, Roman D.
Professor, Univ. of Michigan, Dept. of Civil and Environmental Engineering, Ann Arbor, MI
(Received 27 November 2012; accepted 22 March 2013)
An image-based method has been developed for rapidly determining the size distribution of coarse sand and gravel. The system utilizes a back-lit tilting “translucent segregation table” (TST) to segregate the particles by size prior to photographing the specimen from above. Most significantly, the soil particles need not to be detached from each other; they only need to rest on the translucent plate in a single layer. An image-processing method called “watershed segmentation” digitally separates the soil particles so that each one is individually accounted for in the resulting size distribution. A sieve-equivalent particle size, which considers the passage of ellipsoids through square sieve openings, is obtained for every particle in the specimen. Twenty soil specimens were tested by the TST and by sieving. Slight differences between sieve and TST results were observed and explanations for the minor discrepancies were explored. Considering that the TST uses a single camera and just one photograph, it appears to be an attractive clean, rapid, and low-cost alternative to sieving.
Paper ID: GTJ20120221