论文信息 - Kanji Character Detection from Complex Real Scene Images based on Character Properties

Kanji Character Detection from Complex Real Scene Images based on Character Properties

Character recognition in complex real scene images is a very challenging undertaking. The most popular approach is to segment the text area using some extra pre-knowledge, such as "characters are in a signboard'', etc. This approach makes it possible to construct a very time-consuming method, but generality is still a problem. In this paper, we propose a more general method by utilizing only character features. Our algorithm consists of five steps: pre-processing to extract connected components, initial classification using primitive rules, strong classification using AdaBoost, Markov random field (MRF) clustering to combine connected components with similar properties, and post-processing using optical character recognition (OCR) results. The results of experiments using 11 images containing 1691 characters (including characters in bad condition) indicated the effectiveness of the proposed system, namely, that 52.9% of characters were extracted correctly with 625 noise components extracted as characters.

Hiroshi Sako | Hiroto Nagayoshi | Lianli Xu

[1] King Ngi Ngan,et al. A robust Markovian segmentation based on highest confidence first (HCF) , 1997, Proceedings of International Conference on Image Processing.

[2] Yoav Freund,et al. A decision-theoretic generalization of on-line learning and an application to boosting , 1997, EuroCOLT.

[3] Shigeru Akamatsu,et al. Recognizing Characters in Scene Images , 1994, IEEE Trans. Pattern Anal. Mach. Intell..

[4] Li Xu,et al. Automatic character detection and segmentation in natural scene images , 2007 .

[5] Yoav Freund,et al. A decision-theoretic generalization of on-line learning and an application to boosting , 1995, EuroCOLT.

[6] Hao Wang,et al. Automatic character location and segmentation in color scene images , 2001, Proceedings 11th International Conference on Image Analysis and Processing.

[7] Cheng-Lin Liu,et al. Text/Non-text Ink Stroke Classification in Japanese Handwriting Based on Markov Random Fields , 2007 .

[8] Milan Sonka,et al. Image processing analysis and machine vision [2nd ed.] , 1999 .

[9] Jia Zeng,et al. Markov Random Field-Based Statistical Character Structure Modeling for Handwritten Chinese Character Recognition , 2008, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[10] P. Pérez,et al. Markov random fields and images , 1998 .

[11] Xilin Chen,et al. Automatic detection and recognition of signs from natural scenes , 2004, IEEE Transactions on Image Processing.