A Two-step Character Binarization Method Using Standard Deviationin Local Area
2
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
Generally, non-uniform illumination and ambiguous intensity make much effects on binarization results in character extraction. Especially. in case of character extraction on a non-rigid surface, the performance become worse because characters have changeful shape, illumination, and intensity. In the paper, a two-step approach across a non-rigid surface is proposed. This method is composed of a character extraction step and a character verification step. A two-directions standard deviation algorithm and a modified binarization method are used to segment the character’s area and determine character’s position in the first step. In this step, our adaptive thresholding is strengthened by our two directions standard deviation algorithm when the characters is separated from its background and selects the best threshold value of each pixel by the information of local maxima and minima of intensity value in the local window. In the second step, the extracted characters are verified with given conditions such as width, height, and distribution of a character. The proposed binarization algorithm demonstrates superior performance in extracting characters among the non-rigid object with various conditions after performing extensive experiments.
Abstract
Generally, non-uniform illumination and ambiguous intensity make much effects on binarization results in character extraction. Especially. in case of character extraction on a non-rigid surface, the performance become worse because characters have changeful shape, illumination, and intensity. In the paper, a two-step approach across a non-rigid surface is proposed. This method is composed of a character extraction step and a character verification step. A two-directions standard deviation algorithm and a modified binarization method are used to segment the character’s area and determine character’s position in the first step. In this step, our adaptive thresholding is strengthened by our two directions standard deviation algorithm when the characters is separated from its background and selects the best threshold value of each pixel by the information of local maxima and minima of intensity value in the local window. In the second step, the extracted characters are verified with given conditions such as width, height, and distribution of a character. The proposed binarization algorithm demonstrates superior performance in extracting characters among the non-rigid object with various conditions after performing extensive experiments.
참고문헌 (11)
- X. Zhang, "A Combined Algorithm for Video Text Extraction", 2010 Seventh International Conference on Fuzzy Systems and Knowledge Discovery (2010)
- R. F. Moghaddam, "Adotsu: An adaptive and parameterless generalization of Otsu’s method for document image binarization" (2011)[DOI]
- Robust Car License Plate Localization using a Novel Texture Descriptor, "C. D. Nguyen", Advanced Video and Signal Based Surveillance, ~ (2009)
- J.-C. Wu, "Morphology-based text line extraction", Machine Vision and Applications, 19, pp.195~207 (2008)[DOI]
- Y. Wang, "Adaptive Binarization: A New Approach to License Plate Characters Segmentation", Proc. of International Conference Wavelet Analysis and Pattern Recognition (2012)
- K. W. Jeon, "Extraction of Multiple Car License Plate Areas Using HSI Color Model and Morpological Characteristics", Journal of Korean Institute of Genration Computing, 9(2), pp.59~63 (2009)
- J. Duan, "A Multi-stage Adaptive Binarization Scheme for Document Images", International Joint Conference on Computational Sciences and Optimization (2009)
- N. Otsu, "A Threshold Selection Method from Gray-level Histogram", IEEE Trans. Syst. Man Cybern, 9, pp.62~66 (1979)
- J. Bernsen, "Dynamic Thresholding of Graylevel Images", Proc.of 8th International Conference Pattern Recognition (1986)
- W. Niblack, "An Introduction to Digital Image Processing" (1986), Prentice Hall
- R. Fisher, "Hypermedia Image Processing Reference" (1996), Wiley