期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2016
卷号：5
期号：7
页码：13160
DOI：10.15680/IJIRSET.2016.0507190
出版社：S&S Publications
摘要：Scene text recognition has inspired great interests from the computer vision community in recent years.We propose a novel scene text recognition method integrating structure guided character detection and linguisticknowledge. We use part-based tree structure to model each category of characters so as to detect and recognizecharacters simultaneously. Since the character models make use of both the local appearance and global structureinformation’s, the detection results are more reliable. For word recognition, we combine the detection scores andlanguage model into the posterior probability of character sequence from the Bayesian decision view. The final wordrecognition result is obtained by maximizing the character sequence posterior probability via Viterbi algorithm.Experimental results on a range of challenging public data sets demonstrate that the proposed method achieves state-ofthe-art performance both for character detection and word recognition.
关键词：Character recognition; Scene text recognition; Part-Based Tree-Structured Models (TSMs); Posterior;probability; Word recognition
