20. Vision by Graph Pyramids

Abstract To efficiently process huge amounts of structured sensory data for vision, graph pyramids are pro­ posed. Hierarchies of graphs can be generated by dual graph contraction. The goal is to reduce the data structure by a constant reduction factor while preserving certain image properties, like connectivity. While implemented versions solve several technical vision problems like image seg­ mentation, the framework can be used as a model for biological systems, too.

[1]  G. Jones,et al.  Theory of Maps on Orientable Surfaces , 1978 .

[2]  Edward H. Adelson,et al.  The Laplacian Pyramid as a Compact Image Code , 1983, IEEE Trans. Commun..

[3]  Azriel Rosenfeld,et al.  Multiresolution image processing and analysis , 1984 .

[4]  Leonard Uhr,et al.  The Several Steps from Icon to Symbol Using Structured Cone/Pyramids , 1984 .

[5]  Pascal Lienhardt,et al.  Subdivisions of n-dimensional spaces and n-dimensional generalized maps , 1989, SCG '89.

[6]  Azriel Rosenfeld,et al.  A critical view of pyramid segmentation algorithms , 1990, Pattern Recognit. Lett..

[7]  M. N. Shanmukha Swamy,et al.  Graphs - theory and algorithms , 1992 .

[8]  Azriel Rosenfeld,et al.  A Pyramid Framework for Early Vision , 1994 .

[9]  W. Kropatsch.  Building irregular pyramids by dual-graph contraction , 1995 .

[10]  Walter G. Kropatsch,et al.  Universal segmentation with PIRRAMIDS , 1996 .

[11]  Walter G. Kropatsch,et al.  Equivalent contraction kernels to build dual irregular pyramids , 1996, TFCV.

[12]  A. Rosenfeld,et al.  Curve Detection in a Noisy Image , 1997, Vision Research.

[13]  Azriel Rosenfeld,et al.  Local deformations of digital curves , 1997, Pattern Recognit. Lett..

[14]  Jean-Michel Jolion,et al.  Object matching on irregular pyramid , 1998, Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170).

[15]  Walter G. Kropatsch,et al.  A Minimal Line Property Preserving Representation of Line Images , 1999, Computing.