Authors:
Tzu C. Shen
and
Andrés R. Guesalaga
Affiliation:
Catholic University of Chile, Chile
Keyword(s):
Map-matching, pattern recognition, radar calibration.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Feature Extraction
;
Features Extraction
;
Image and Video Analysis
;
Informatics in Control, Automation and Robotics
;
Signal Processing, Sensors, Systems Modeling and Control
Abstract:
This paper describes a new method of image pattern recognition based on the Hausdorff Distance. The
technique looks for similarities between a given pattern and its possible representations within an image. This method performs satisfactorily when confronted to image perturbations or partial occlusions. An extension of the classical Hausdorff Distance technique chooses the best candidate among multiple suboptimal solutions. The search strategy is based on the Branch and Bounds algorithm, where cells with low probability of containing the optimal solution are pruned, while feasible cells are divided again until the optimal solution is found. By using this strategy, exhaustive and no-informative searches are avoided among the possible combinations, reducing the processing time considerably. A case study is presented, where the proposed method is applied to calibration of surveillance radars using hydrographic charts as models for the radar echo images.