期刊名称：International Journal of Engineering and Computer Science
印刷版ISSN：2319-7242
出版年度：2015
卷号：4
期号：5
页码：11797-11803
出版社：IJECS
摘要：Detecting text in natural images is an important prerequisite. Text Characters and Strings provide valuable information for manyapplications, extracting text directly from natural scene images or videos is a challenging task because of diverse text patterns and variousbackground interferences. In this paper an attempt is made to extract text in Natural Scene image. This text detection algorithm employsedge-Enhanced Maximally Stable Extremal Regions as basic letter candidates. These candidates are then filtered using geometric and strokewidth information to exclude non-text objects. Letters are paired to identify text lines, which are subsequently separated into words. Thecoding has implemented in MATLAB R2014a and the qualitative analysis is done. Based on the simplicity of eliminating clutteredbackground. we followed other techniques in extracting text easily such as color thresholding app and Region of interest (ROI).According tothe statements in text extracted from natural images the controlling of electrical motors can also be done. This can be explained by hardwareimplementation of text extraction application using ARDUINO kit.
关键词：Edge Enhanced MSER; Color Thresholding;Region of interest (ROI); Text extraction; Image processing
