Share Email Print

Proceedings Paper

Probabilistic model of multiple dynamic curve matching for a semitransparent scene
Author(s): Hail Mallouche; Jacques A. de Guise; Yves Goussard
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

Active contours have become an attractive subject in computer vision. Connectivity and closure properties of these contours help to overcome some important difficulties in computer vision, (1) edge organization and (2) region merging. Consequently, using deformable contours brings additional knowledge in the form of curve properties, and thus reduces dramatically the search space dimension. Many researchers described a dynamic contour as a snake curve which is deformed so as to minimize the curvilinear integral of the sum of internal and external local energies. Here to fore, proposals were restricted to using a single deformable curve. In this paper, we present a probabilistic model of multiple dynamic non- parametric curve matching with 2D images. The model is made up of two parts: (1) image formation, and (2) high-level interactions, which measure consistency of the reconstructed object with the given image and the relational a priori information of the object, respectively. The model is conceived for a semitransparent scene such as additive projection imaging - planar radiographical transmission or planar emission imaging. The scene model represents a hierarchical structure of three processes: lines, regions and relational graphs. Thus, an object is modeled as a set of linked subobjects according to a 3D relational graph which can be projected from a known viewpoint as a 2D region relational graph. The graph represents high- level structural knowledge which reduces the search space dimension by conditioning the a posteriori probability to the a priori known graph. The resultant objective function is optimized using a descending search method with randomized sampling. Finally, successful results are presented for object matching in semitransparent noisy synthetic scenes.

Paper Details

Date Published: 11 August 1995
PDF: 10 pages
Proc. SPIE 2573, Vision Geometry IV, (11 August 1995); doi: 10.1117/12.216410
Show Author Affiliations
Hail Mallouche, Ecole Polytechnique de Montreal (Canada)
Jacques A. de Guise, Ecole de Technologie Superieure (Canada)
Yves Goussard, Ecole Polytechnique de Montreal (Canada)

Published in SPIE Proceedings Vol. 2573:
Vision Geometry IV
Robert A. Melter; Angela Y. Wu; Fred L. Bookstein; William D. K. Green, Editor(s)

© SPIE. Terms of Use
Back to Top