Table Structure Extraction with Bi-Directional Gated Recurrent Unit Networks

Tables present summarized and structured information to the reader, which makes table's structure extraction an important part of document understanding applications. However, table structure identification is a hard problem not only because of the large variation in the table layouts and styles, but also owing to the variations in the page layouts and the noise contamination levels. A lot of research has been done to identify table structure, most of which is based on applying heuristics with the aid of optical character recognition (OCR) to hand pick layout features of the tables. These methods fail to generalize well because of the variations in the table layouts and the errors generated by OCR. In this paper, we have proposed a robust deep learning based approach to extract rows and columns from a detected table in document images with a high precision. In the proposed solution, the table images are first pre-processed and then fed to a bi-directional Recurrent Neural Network with Gated Recurrent Units (GRU) followed by a fully-connected layer with softmax activation. The network scans the images from top-to-bottom as well as left-to-right and classifies each input as either a row-separator or a column-separator. We have benchmarked our system on publicly available UNLV as well as ICDAR 2013 datasets on which it outperformed the state-of-theart table structure extraction systems by a significant margin.

[1]  Kuldip K. Paliwal,et al.  Bidirectional recurrent neural networks , 1997, IEEE Trans. Signal Process..

[2]  Yoshua Bengio,et al.  Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling , 2014, ArXiv.

[3]  Thomas M. Breuel,et al.  Performance Evaluation and Benchmarking of Six-Page Segmentation Algorithms , 2008, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  Thomas Kieninger,et al.  An open approach towards the benchmarking of table structure recognition systems , 2010, DAS '10.

[5]  Thomas Kieninger,et al.  The T-Recs Table Recognition and Analysis System , 1998, Document Analysis Systems.

[6]  Daniel P. Lopresti,et al.  Table structure recognition and its evaluation , 2000, IS&T/SPIE Electronic Imaging.

[7]  Stephan Lewandowsky,et al.  The Perception of Statistical Graphs , 1989 .

[8]  Andreas Dengel,et al.  DeCNT: Deep Deformable CNN for Table Detection , 2018, IEEE Access.

[9]  Abdel Belaïd,et al.  Table information extraction and structure recognition using query patterns , 2015, 2015 13th International Conference on Document Analysis and Recognition (ICDAR).

[10]  H.S. Baird,et al.  A retargetable table reader , 1997, Proceedings of the Fourth International Conference on Document Analysis and Recognition.

[11]  Andreas Dengel,et al.  DeepDeSRT: Deep Learning for Detection and Structure Recognition of Tables in Document Images , 2017, 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR).

[12]  J. Cordy,et al.  A Survey of Table Recognition : Models , Observations , Transformations , and Inferences , 2003 .

[13]  Thomas Kieninger,et al.  Applying the T-Recs table recognition system to the business letter domain , 2001, Proceedings of Sixth International Conference on Document Analysis and Recognition.

[14]  Alexey O. Shigarov,et al.  Configurable Table Structure Recognition in Untagged PDF documents , 2016, DocEng.

[15]  Muhammad Imran Malik,et al.  Table Detection Using Deep Learning , 2017, 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR).

[16]  Thomas M. Breuel,et al.  Efficient implementation of local adaptive thresholding techniques using integral images , 2008, Electronic Imaging.

[17]  Thomas Kieninger,et al.  Table Recognition and Labeling Using Intrinsic Layout Features , 1999 .

[18]  Yalin Wang,et al.  Table structure understanding and its performance evaluation , 2004, Pattern Recognit..

[19]  Yalin Wang,et al.  Automatic table ground truth generation and a background-analysis-based table structure extraction method , 2001, Proceedings of Sixth International Conference on Document Analysis and Recognition.