An Efficient Method for Text Location and Segmentation
Authors:
- Weijuan WenWWWeijuan Wen
- Xianglin HuangXHXianglin Huang
- Lifang YangLYLifang Yang
- Zhao YangZYZhao Yang
- Pengju ZhangPZPengju Zhang
- 1 January 2009
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3, 3-7
Abstract
Recently, with the development of the information technology, the number of images is increasing rapidly. There is much important text in these images. In order to satisfy the need of images viewing, text identification and text retrieval, this paper presents an efficient Method for text location and segmentation. Firstly, the image is converted into wavelet transform domain. Then three sub-band wavelet images of HxLy, LxHy, HxHy are binarizated and merged into a texture image by linear combination. The texture image is processed by CRLA(constrained run length algorithm) and image smoothing to enhance the candidate text regions. Finally, the text regions are located after 8-connected component growth and filtering out non-text regions. In order to be recognized by OCR(Optical Character Recognition), characters are segmented from text blocks based on the mean values of some slices of text-block. Experimental results show that the proposed method is robust to overlapped complex background.Keywords
This publication has 4 references indexed in Scilit:
- Document image segmentation using wavelet scale-space featuresIEEE Transactions on Circuits and Systems for Video Technology, 2002
- Automatic text extraction from color image2000
- A theory for multiresolution signal decomposition: the wavelet representationIEEE Transactions on Pattern Analysis and Machine Intelligence, 1989
- Block segmentation and text extraction in mixed text/image documentsComputer Graphics and Image Processing, 1982