سال انتشار: ۱۳۸۹

محل انتشار: سومین همایش ملی مهندسی کامپیوتر و فناوری اطلاعات

تعداد صفحات: ۴

نویسنده(ها):

davod zaravi – Islamic azad university of shushtar
Iman urak – Islamic azad university of izeh
Iman dahhar – Jahad daneshgahi institut of ahvaz

چکیده:

Text extraction is one of most common topics in recent researches. It can be used for indexing images or search in data.Most existing hard copies in libraries could be digitally. It is pre processing for OCR .In this paper a new robust and efficient algorithm is proposed to automatic text extraction from colored book and journal cover sheets. First, wavelet transform performs. Next, a dynamic threshold is used for edge detecting from detail wavelet coefficient. By blurring approximate coefficients with alternative heuristic thresholding, achieve effective edges. Afterward, with ROI technique gets binary image. Finally text boxes would be extracted with new projection profile.