Facet model and mathematical morphology for surface characterization

Besma R. Abidi, Hamed Sari-Sarraf, James S. Goddard, Martin A. Hunt

Research output: Contribution to journalConference articlepeer-review

2 Scopus citations

Abstract

This paper describes an algorithm for the automatic segmentation and representation of surface structures and non-uniformities in an industrial setting. The automatic image processing and analysis algorithm is developed as part of a complete on-line web characterization system of a papermaking process at the wet end. The goal is to: (1) link certain types of structures on the surface of the web to known machine parameter values, and (2) find the connection between detected structures at the beginning of the line and defects seen on the final product. Images of the pulp mixture (slurry), carried by a fast moving table, are obtained using a stroboscopic light and a CCD camera. This characterization algorithm succeeded where conventional contrast and edge detection techniques failed due to a poorly controlled environment. The images obtained have poor contrast and contain noise caused by a variety of sources. After a number of enhancement steps, conventional segmentation methods still failed to detect any structures and are consequently discarded. Techniques tried include the Canny edge detector, the Sobel, Roberts, and Prewitt's filters, as well as zero crossings. The facet model algorithm, is then applied to the images with various parameter settings and is found to be successful in detecting the various topographic characteristics of the surface of the slurry. Pertinent topographic elements are retained and a filtered image computed. Carefully tailored morphological operators are then applied to detect and segment regions of interest. Those regions are then selected according to their size, elongation, and orientation. Their bounding rectangles are computed and represented. Also addressed in this paper are aspects of the real time implementation of this algorithm for on-line use. The algorithm is tested on over 500 images of slurry and is found to segment and characterize nonuniformities on all 500 images.

Original languageEnglish
Pages (from-to)334-344
Number of pages11
JournalProceedings of SPIE - The International Society for Optical Engineering
Volume3837
StatePublished - 1999
EventProceedings of the 1999 Intelligent Robots and Computer Vision XVIII: Algorithms, Techniques, and Active Vision - Boston, MA, USA
Duration: Sep 20 1999Sep 21 1999

Fingerprint Dive into the research topics of 'Facet model and mathematical morphology for surface characterization'. Together they form a unique fingerprint.

Cite this