Text Localization and Extraction in Natural Scene Images
International Journal on Recent and Innovation Trends in Computing and Communication, 3 (2):
512--516 (February 2015)DOI: 10.17762/ijritcc2321-8169.150218
Abstract
Content based image analysis methods are receiving more attention in recent time due to increase in use of image capturing devices. Among all contents in images, text data finds wide applications such as license plate reading, mobile text recognition and so on. The Text Information Extraction (TIE) is a process of extraction of text from images which consists of four stages: detection, localization, extraction and recognition. Text detection and localization plays important role in system’s performance. The existing methods for detection and localization, region based and connected component based, have some limitations due difference in size, style, orientation etc. To overcome the limitations, a hybrid approach is proposed to detect and localize text in natural scene images. This approach includes steps: pre-processing connected component analysis, text extraction