Citation link:
Authors:
Венков, Пенчо Георгиев; Venkov, Pentcho Georgiev
Title:
Моделно-воден анализ на изображения
Other Titles:
Model-guided analysis of grey level images
Date of Issue:
1995
Is Part of:
Proceedings of the Technical University – Sofia, 48(4), 1995, pp. 379-388
Publisher:
Технически университет - София
Identifiers:
0374-342X [issn]
Type:
Article
Language:
bul
Subject:
плоски обекти; кон¬турни сегменти; двумерни контурни модели; алгоритми за генериране на хипотези
Abstract:
A method is developed to recognize and locate randomly oriented and partially occulted "flat" objects. It is based upon matching the edge descriptions of an image with models of 2 D objects, represented by segmented polygonal contours. The method uses a technique of hypotheses generation to predict the position of a model in the scene by matching privileged segments of the model to compatible segments in the image. An iterative algorithm for hypotheses evaluation takes advantage of the predicted position of the model to identify additional segments in the image and to update the model position. The matching ends with a final position estimate of the recognized object in the scene if a high quality measure is reached by a hypo these or when a limited number of the best first hypotheses have been evaluated and rejected. The recognition method has been integrated within a vision system for industrial robot and tested on a number of different tasks.
Description:
Съдържа формули. – Библиогр., с. 388.
