GENERATION AND CODIFICATION OF CONNECTED COMPONENTS IN IMAGES OF PIXEL DIAGNOSIS

A methodology is introduced for the generation of a primary segmentation of a scene image and extraction of its connected components from several diagnostic maps. The segmentation is based on an adaptation of the Watershed transform. Also presented is an extension of the algorithm described in Mandler (1990) for the analysis of the connected components contained in the color map produced by the Watershed transform. The algorithm performs the connected components analysis, returning as results every component in the color map, its border code, and an ordered list of its neighbors including the coordinates of the first common border point. This is done efficiently, scanning every line in the color map only once. The original algorithm has been extended to deal with 8-connectivity.

[1]  Juan Méndez,et al.  A Model for a Structural Vision System , 1989, EUROCAST.

[2]  Luc Vincent,et al.  Watersheds in Digital Spaces: An Efficient Algorithm Based on Immersion Simulations , 1991, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Martin D. Levine,et al.  Low Level Image Segmentation: An Expert System , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  Serge Beucher,et al.  Use of watersheds in contour detection , 1979 .

[5]  Heinrich Niemann Pattern Analysis and Understanding , 1990 .

[6]  Eberhard Mandler,et al.  One-pass encoding of connected components in multivalued images , 1990, [1990] Proceedings. 10th International Conference on Pattern Recognition.

[7]  Serge Beucher,et al.  Segmentation tools in mathematical morphology , 1990, Optics & Photonics.

[8]  T. Pavlidis Algorithms for Graphics and Image Processing , 1981, Springer Berlin Heidelberg.

[9]  F. M. Hernández,et al.  DEVELOPMENT TOOL FOR COMPUTER VISION SYSTEMS AT PIXEL LEVEL , 1994 .

[10]  Ramesh C. Jain,et al.  Knowledge representation and control in computer vision systems , 1988, IEEE Expert.