Extraction and recognition of multi-oriented text from trademark images

This paper presents an application system to extract and recognize text in Trademark and Logo images. Many methods have been proposed to address image text detection and recognition problems, and some of them have achieved impressive results for some specific applications. But recognition of Multi-oriented characters and numerals in Trademark and Logo images is still a challenge due to the variations of text font, size, color, orientation and alignment, and it is often affected by complex background and non-textual objects. Also to prevent the trademark infringement cases and for an efficient trademark similarity matching system, accurate text recognition plays a vital role.