文章基本信息

标题：Driving Active Contours to Concave Regions
本地全文：下载
作者：Qusay O. Mosa ; Ali Saeed Alfoudi ; Ahmed A. Brisam 等
期刊名称：Webology
印刷版ISSN：1735-188X
出版年度：2022
卷号：19
期号：1
页码：5131-5140
DOI：10.14704/WEB/V19I1/WEB19345
语种：English
出版社：University of Tehran
摘要：Broken characters restoration represents the major challenge of optical character recognition (OCR). Active contours, which have been used successfully to restore ancient documents with high degradations have drawback in restoring characters with deep concavity boundaries. Deep concavity problem represents the main obstacle, which has prevented Gradient Vector Flow active contour in converge to objects with complex concavity boundaries. In this paper, we proposed a technique to enhance (GVF) active contour using particle swarm optimization (PSO) through directing snake points (snaxels) toward correct positions into deep concavity boundaries of broken characters by comparing with genetic algorithms as an optimization method. Our experimental results showed that particle swarm optimization outperform on genetic algorithm to correct capturing the converged areas and save spent time in optimization process.
关键词：GVF;Particle Swarm Optimizer;Deep Concavity