Authors:
Noppon Lertchuwongsa
;
Michèle Gouiffès
and
Bertrand Zavidovique
Affiliation:
IEF, Institut d’ Electronique Fondamentale, Université Paris Sud 11, France
Keyword(s):
Computer Vision, Stereovision, Color Lines, Hausdorff, Shape Matching.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Feature Extraction
;
Features Extraction
;
Image and Video Analysis
;
Image Processing
;
Informatics in Control, Automation and Robotics
;
Robotics and Automation
;
Signal Processing, Sensors, Systems Modeling and Control
;
Vision, Recognition and Reconstruction
Abstract:
Level lines and sets are competitive features to support recognition. Color is assumed more informative than intensity, so color-lines are preferred to exhibit the set basis. In the paper, they are defined after level lines, and then extracted and characterize. Greater information is kept by color lines resulting into more efficient grouping towards objects. A novel Hausdorff-inspired disparity finder is introduced fed in by color lines with respect to epipolar constraints. The efficient disparity map resulting from pixel wise line-matching between left and right images justifies our technical choices.