Title :
Edge based texture segmentation
Author :
Vaidyanathan, G. ; Lynch, P.M.
Author_Institution :
Dept. of Mech. Eng., New Orleans Univ., LA, USA
Abstract :
A method for detecting edges in a multitextural image is proposed. The edge-based procedure uses a texture edge operator in a single-pass algorithm. The operator is based on the differences of mean texture measures extracted from a set of adjacent image neighborhood regions. The absolute mean of the set of regions are evaluated. To discriminate between points which are edges of texture elements and points which are far from edges, a tolerance value is imposed on the absolute mean deviations. Deviations greater than the tolerance value are taken to indicate an edge. The technique is shown to be effective with natural and artificial textures
Keywords :
picture processing; absolute mean deviations; artificial textures; boundary detection; edge based texture segmentation; multitextural image; natural textures; single-pass algorithm; texture edge operator; tolerance value; Data mining; Image edge detection; Image segmentation; Image texture analysis; Joining processes; Layout; Mechanical engineering; Merging; Pixel;
Conference_Titel :
Southeastcon '90. Proceedings., IEEE
Conference_Location :
New Orleans, LA
DOI :
10.1109/SECON.1990.117993
