SÜPERPİKSEL TABANLI SATIR BÖLÜTLEME

Satır bölütleme tarihi doküman analizi uygulamalarının en temel aşamalarından birisidir. Satır bölütleme başarısı, daha sonraki doküman analizi yöntemlerinin başarısını doğrudan etkilemektedir. Matbu belgelerde kayıpsız satır bölütleme işlemi kolaylıkla yapılabilmektedir. Ancak, el yazımı belgeler için satır bölütleme işlemi metin satırlarının eğik, eğri, dalgalı olması, satırlar arası boşlukların darlığı, örtüşen ve temas eden bileşenlerden dolayı hala zorlayıcı bir problemdir. Bu çalışmada, el yazımı dokümanlar için süperpiksel tabanlı yeni bir satır bölütleme yöntemi önerilmiştir. Yöntem ardışık satırları bölütleyebilen en güvenli sınırın elde edilmesini hedeflemektedir. Önerilen yöntem 853 adet Çince el yazımı doküman imgesi içeren HIT-MW veri seti üzerinde uygulanmıştır. Veri setinin en önemli özelliği eğik, temas eden ve örtüşen satır davranışlarına sahip imgelerden oluşmasıdır. Önerilen yöntem ile % 98.03 tespit oranı, % 97.66 tanıma doğruluğu elde edilmiş ve yöntemin başarısı literatürde bulunan diğer yöntemlerle karşılaştırılmıştır. Elde edilen sonuçlar ışığında önerilen yöntemin el yazımı metinlerde satır bölütleme uygulamalarındaki başarısı ve potansiyeli ortaya konmuştur.

Anahtar Kelimeler:

Süperpiksel, Satır Bölütleme, El Yazısı Belge, Doküman İmgesi Analizi, Metin Satırı Çıkarımı, Havza Bölütleme

SUPERPIXEL BASED TEXT LINE SEGMENTATION

Text line segmentation is one of the essential stages of historical document analysis applications. The accuracy of text line segmentation affects directly the success of following document analysis steps. For printed documents, lossless text line segmentation can be done readily. But, for handwritten documents, unfortunately it is still a challenging problem because of the skewed, curved, fluctuated text lines, narrow gaps between the text lines, overlapping and touching components. In this paper, a novel superpixel-based text line segmentation method for handwritten documents is proposed. This method aims to extract the most reliable boundary to segment consecutive text lines. This method is implemented HIT-MW dataset containing 853 Chinese handwritten document images. The most important feature of this dataset is to be composed of documents having skewed, overlapping and touching text lines. A detection rate of 98.03% and a recognition accuracy of 97.66% is obtained and these results are compared with the ones of existing state of the art methods. With these results, segmentation success and potential of our method for handwriting text line segmentation is pointed out.

Keywords:

Superpixel, Text Line Segmentation, Handwritten Document, Document Image Analysis, Watershed Segmentation,

PDF

___

Adiguzel, H., Sahin, E., Duygulu, P., 2012. A Hybrid Approach for Line Segmentation in Handwritten Documents. In Frontiers in Handwriting Recognition (ICFHR), 2012 International Conference on (pp. 503-508). IEEE.
Alaei, A., Pal, U., Nagabhushan, P., 2011. A New Scheme for Unconstrained Handwritten Text-Line Segmentation. Pattern Recognition, 44(4), 917-928.
Arivazhagan, M., Srinivasan, H., Srihari, S. (2007, January). A statistical approach to line segmentation in handwritten documents. In Document Recognition and Retrieval XIV (Vol. 6500, p. 65000T). International Society for Optics and Photonics.
Ataer, E., Duygulu, P., 2006. Retrieval of Ottoman Documents. In Proceedings of the 8th ACM International Workshop on Multimedia Information Retrieval (MIR '06). ACM, New York, NY, USA, 155-162.
Du, X., Pan, W., Bui, T. D. (2009). Text line segmentation in handwritten documents using Mumford–Shah model. Pattern Recognition, 42(12), 3136-3145.
Fulkerson, B., Vedaldi, A., Soatto, S. (2009, September). Class segmentation and object localization with superpixel neighborhoods. In Computer Vision, 2009 IEEE 12th International Conference on (pp. 670-677). IEEE.
Han, X., Yao, H., Zhong, G. (2017, February). Handwritten text line segmentation by spectral clustering. In Eighth International Conference on Graphic and Image Processing (ICGIP 2016) (Vol. 10225, p. 102251A). International Society for Optics and Photonics.
He, X., Zemel, R. S., Ray, D. (2006, May). Learning and incorporating top-down cues in image segmentation. In European conference on computer vision (pp. 338-351). Springer, Berlin, Heidelberg.
Hoiem, D., Efros, A. A., Hebert, M. (2005, July). Automatic photo pop-up. In ACM transactions on graphics (TOG) (Vol. 24, No. 3, pp. 577-584). ACM.
Hull, J. J. (1998). Document Image Skew Detection: Survey and Annotated Bibliography. Document Analysis Systems II, 40-64.
Koo, H. I., Cho, N. I. (2012). Text-line Extraction in Handwritten Chinese Documents Based on An Energy Minimization Framework. IEEE Transactions on Image Processing, 21(3), 1169-1175.
Li, Y., Sun, J., Tang, C. K., Shum, H. Y. (2004, August). Lazy snapping. In ACM Transactions on Graphics (ToG) (Vol. 23, No. 3, pp. 303-308). ACM.
Li, Y., Zheng, Y., Doermann, D., Jaeger, S. (2008). Script-Independent Text Line Segmentation in Freestyle Handwritten Documents. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(8), 1313-1329.
Likforman-Sulem, L., Hanimyan, A., Faure, C. (1995, August). A Hough Based Algorithm for Extracting Text Lines in Handwritten Documents. In Document Analysis and Recognition, 1995., Proceedings of the Third International Conference on (Vol. 2, pp. 774-777). IEEE.
Louloudis, G., Gatos, B., Pratikakis, I., Halatsis, K. (2006, October). A Block-Based Hough Transform Mapping for Text Line Detection in Handwritten Documents. In Tenth International Workshop on Frontiers in Handwriting Recognition. Suvisoft.
Louloudis, G., Gatos, B., Pratikakis, I., Halatsis, C. (2008). Text Line Detection in Handwritten Documents. Pattern Recognition, 41(12), 3758-3772.
Nagy, G., Seth, S., Viswanathan, M. (1992). A prototype document image analysis system for technical journals. Computer, 25(7), 10-22.
O'Gorman, L. (1993). The document spectrum for page layout analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(11), 1162-1173.
Ryu, J., Koo, H. I., Cho, N. I. (2014). Language-Independent Text-Line Extraction Algorithm for Handwritten Documents. IEEE Signal Processing Letters, 21(9), 1115-1119.
Saabni, R., Asi, A., El-Sana, J. (2014). Text Line Extraction for Historical Document Images. Pattern Recognition Letters, 35, 23-33.
Soille, P. (2013). Morphological Image Analysis: Principles and Applications. Springer Science & Business Media.
Stamatopoulos, N., Gatos, B., Louloudis, G., Pal, U., Alaei, A. (2013, August). ICDAR 2013 handwriting segmentation contest. In Document Analysis and Recognition (ICDAR), 2013 12th International Conference on (pp. 1402-1406). IEEE.
Su, T., Zhang, T., Guan, D. (2007). Corpus-based HIT-MW database for offline recognition of general-purpose Chinese handwritten text. International Journal of Document Analysis and Recognition (IJDAR), 10(1), 27.
Vincent, L., Soille, P. (1991). Watersheds in digital spaces: an efficient algorithm based on immersion simulations. IEEE Transactions on Pattern Analysis & Machine Intelligence, (6), 583-598.
Yin, F., Liu, C. L. (2009). Handwritten Chinese text line segmentation by clustering with distance metric learning. Pattern Recognition, 42(12), 3146-3157.
Zhang, L., Gao, Y., Xia, Y., Lu, K., Shen, J., Ji, R. (2014). Representative Discovery of Structure Cues for Weakly-Supervised Image Segmentation. IEEE Trans. Multimedia, 16(2), 470-479.
Ziaratban, M., Faez, K. (2010, August). An adaptive script-independent block-based text line extraction. In Pattern Recognition (ICPR), 2010 20th International Conference on (pp. 249-252). IEEE.