Texture‐based image segmentation applied to the quantification of superficial sand in salmonid river gravels

The presence of ?ne sediment in river gravels is widely recognized as being detrimental to salmonid habitat quality. In order to facilitate quanti?cation of sand presence at larger scales, this paper presents an application of image processing allowing for rapid and accurate assessments of super?cial sand presence in dry exposed ?uvial gravels. Images for the process are acquired with a 35 mm SLR ?lm camera and then scanned with a desktop scanner. Texture‐based segmentation is then applied to differentiate between sand and clast areas. Results show that the method is accurate and therefore it offers an alternative to bulk sampling in cases where rapid assessments of sand presence are required. Copyright © 2005 John Wiley & Sons, Ltd.