29 March 1988 Segmenting Intersecting And Incomplete Boundaries
Author Affiliations +
This paper describes general purpose algorithms for segmenting boundary images. Intersecting and incomplete boundaries commonly occur in line-art images, and also in natural images depicting translucent objects. Humans have little difficulty in segmenting such boundaries into sets corresponding to the perceptually significant regions in the image. Many existing machine vision algorithms, however, have difficulty in processing images which contain intersecting or incomplete boundaries. However, Walters' segmentation algorithm based on the p-space representation of oriented edges will correctly segment images with non-acutely intersecting lines and boundaries. This paper suggests a non-iterative, parallel algorithm which will fill large gaps and segment acutely intersecting boundaries. These algorithms are useful in a variety of applications including fake color separation, character recognition, and also in segmenting images depicting translucent objects.
© (1988) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Ganapathy Krishnan, Ganapathy Krishnan, Deborah Walters, Deborah Walters, "Segmenting Intersecting And Incomplete Boundaries", Proc. SPIE 0937, Applications of Artificial Intelligence VI, (29 March 1988); doi: 10.1117/12.947019; https://doi.org/10.1117/12.947019


Adaptive edge detection in a global optimal observation scale
Proceedings of SPIE (December 04 2011)
Efficient detection of ellipses from an image by a guided...
Proceedings of SPIE (February 10 2009)
Object pre-processing for optical pattern recognition
Proceedings of SPIE (January 10 2005)
Parallel Edge Detection
Proceedings of SPIE (May 17 1987)

Back to Top