NLP&CC 2012 报告人:许灿辉 单 位:北京大学计算机科学技术研究所 Integration of Text Information and Graphic Composite for PDF Document Analysis 基于复合图文整合的 PDF 文档分析 Integration of.

Slides:



Advertisements
Similar presentations
YEARBOOK Layout and Design.
Advertisements

Chapter 3 – Web Design Tables & Page Layout
ADA Compliant Websites & Documents What the heck am I supposed to do?
McGraw-Hill/Irwin The O’Leary Series © 2002 The McGraw-Hill Companies, Inc. All rights reserved. Microsoft Excel 2002 Lab 2 Charting Worksheet Data.
Carolina Galleguillos, Brian McFee, Serge Belongie, Gert Lanckriet Computer Science and Engineering Department Electrical and Computer Engineering Department.
Reporting Agricultural Research. COMMOM CORE/NEXT GENERATION SCIENCE STANDERD ADDRESSED CCSS.ELA-Literacy.RH Determine the meaning of words and.
Preparing Business Reports
Identifying Image Spam Authorship with a Variable Bin-width Histogram-based Projective Clustering Song Gao, Chengcui Zhang, Wei Bang Chen Department of.
HOW MIGHT WE SUPPORT TEACHERS AS THEY DEEPEN THEIR OWN UNDERSTANDING AND EXPLORE STUDENT THINKING RELATED TO THE K-6 GEOMETRY PROGRESSION? GINI STIMPSON.
Texture Segmentation Based on Voting of Blocks, Bayesian Flooding and Region Merging C. Panagiotakis (1), I. Grinias (2) and G. Tziritas (3)
Word. Define the meaning of Word will be divided into two parts: First Section: What it means is commonly known It is a word processor that through which.
Data Structures For Image Analysis
Hierarchical Region-Based Segmentation by Ratio-Contour Jun Wang April 28, 2004 Course Project of CSCE 790.
Prénom Nom Document Analysis: Segmentation & Layout Analysis Prof. Rolf Ingold, University of Fribourg Master course, spring semester 2008.
2 Part II Enhancing a Presentation Changing the Presentation Design Design template Professionally created slide designs contain –Color schemes –Custom.
Processing Digital Images. Filtering Analysis –Recognition Transmission.
Quadtrees, Octrees and their Applications in Digital Image Processing
LYU 0102 : XML for Interoperable Digital Video Library Recent years, rapid increase in the usage of multimedia information, Recent years, rapid increase.
Chapter 10 Image Segmentation.
CS292 Computational Vision and Language Visual Features - Colour and Texture.
Document Image Analysis CSE 717 An Introduction. Document Image Analysis  DIA is the theory and practice of recovering the symbol structures of digital.
1 Technical Report Writing  Purpose of Report Writing  Structure of Report Writing  Layout of Report Writing.
Common Page Design. Graphics and Tables Uses: Objects Numbers Concepts Words.
DESIGNING DOCUMENTS And page layout. What is document design?  Refers to page layout, that is, where the visuals and information are placed on a page.
ICT Revision. Database – Data Management The insertion and deletion of fields The insertion and deletion of records Tables to be linked together The editing.
CS654: Digital Image Analysis Lecture 3: Data Structure for Image Analysis.
Creating a Document with a Title Page, Lists, Tables, and a Watermark
BACKGROUND LEARNING AND LETTER DETECTION USING TEXTURE WITH PRINCIPAL COMPONENT ANALYSIS (PCA) CIS 601 PROJECT SUMIT BASU FALL 2004.
Chapter 9.  Mathematical morphology: ◦ A useful tool for extracting image components in the representation of region shape.  Boundaries, skeletons,
S EGMENTATION FOR H ANDWRITTEN D OCUMENTS Omar Alaql Fab. 20, 2014.
Organizing Your Information
Enricher Converter Analyzer Parser & Renderer UNIVERSAL, FAST AND RELIABLE.
Word 2010 Vocabulary List 1. Click and Type - A feature that allows you to double-click a blank area of a document to position the cursor in that location,
© Prentice Hall, 2007 Business Communication Essentials, 3eChapter Writing and Completing Reports and Proposals.
Digital Image Processing CCS331 Relationships of Pixel 1.
Chapter 10 Image Segmentation.
McGraw-Hill Career Education© 2008 by the McGraw-Hill Companies, Inc. All Rights Reserved. Office Word 2007 Lab 3 Creating Reports and Tables.
Digital Image Processing Lecture 19: Segmentation: Morphological Watersheds Prof. Charlene Tsai.
Pixel Connectivity Pixel connectivity is a central concept of both edge- and region- based approaches to segmentation The notation of pixel connectivity.
Chapter 4 Working with Frames. Align and distribute objects on a page Stack and layer objects Work with graphics frames Work with text frames Chapter.
A survey of different shape analysis techniques 1 A Survey of Different Shape Analysis Techniques -- Huang Nan.
WORD VOCABULARY LIST #5 MICROSOFT OFFICE WORD VOCABULARY LIST #5 bar chart - A chart with bars that compares the quantities of two or more items.
Levels of Image Data Representation 4.2. Traditional Image Data Structures 4.3. Hierarchical Data Structures Chapter 4 – Data structures for.
© 2010 Delmar, Cengage Learning Chapter 4 Working with Frames.
Accessibility – Standards and Guidelines April 1, 2015.
© 2011 Delmar, Cengage Learning Chapter 4 Working with Frames.
Face Image-Based Gender Recognition Using Complex-Valued Neural Network Instructor :Dr. Dong-Chul Kim Indrani Gorripati.
Colour and Texture. Extract 3-D information Using Vision Extract 3-D information for performing certain tasks such as manipulation, navigation, and recognition.
Low level Computer Vision 1. Thresholding 2. Convolution 3. Morphological Operations 4. Connected Component Extraction 5. Feature Extraction 1.
Digital Image Processing
Scanned Documents INST 734 Module 10 Doug Oard. Agenda Document image retrieval  Representation Retrieval Thanks for David Doermann for most of these.
Positioning Objects with CSS and Tables
Microsoft® Access Generate forms quickly 1 Modify controls in Layout View 2 Work with form sections 3 Modify controls in Design View 4 Add calculated.
Scene Text Extraction Using Focus of Mobile Camera Egyul Kim, SeongHun Lee, JinHyung Kim Artificial Intelligence & Pattern Recognition Lab, KAIST, Korea.
`. Lecture Overview HTML Body Elements Linking techniques HyperText references Linking images Linking to locations on a page Linking to a fragment on.
Machine Vision ENT 273 Hema C.R. Binary Image Processing Lecture 3.
© 2008 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice Niranjan Damera-Venkata HP Labs Design.
Generation of Chinese Character Based on Human Vision and Prior Knowledge of Calligraphy 报告人: 史操 作者: 史操、肖建国、贾文华、许灿辉 单位: 北京大学计算机科学技术研究所 NLP & CC 2012: 基于人类视觉和书法先验知识的汉字自动生成.
1 A Methodology for automatic retrieval of similarly shaped machinable components Mark Ascher - Dept of ECE.
BYST Seg-1 DIP - WS2002: Segmentation Digital Image Processing Image Segmentation Bundit Thipakorn, Ph.D. Computer Engineering Department.
Formal Report Strategies. Types of Formal Reports Informational Presents Info Analytical Presents Info Analyses info and draws conclusions Recommendation.
© 2012 Cengage Learning. All Rights Reserved. This edition is intended for use outside of the U.S. only, with content that may be different from the U.S.
Objectives At the end of this session, students will be able to:
Content-Based Image Retrieval Readings: Chapter 8:
3.1 Clustering Finding a good clustering of the points is a fundamental issue in computing a representative simplicial complex. Mapper does not place any.
Text Detection in Images and Video
ADA Compliant Website & Documents
5.00 Apply procedures to organize content by using Dreamweaver. (22%)
Text Features.
Presentation transcript:

NLP&CC 2012 报告人:许灿辉 单 位:北京大学计算机科学技术研究所 Integration of Text Information and Graphic Composite for PDF Document Analysis 基于复合图文整合的 PDF 文档分析 Integration of Text Information and Graphic Composite for PDF Document Analysis 基于复合图文整合的 PDF 文档分析 2012 年 11 月 04 日

1 、 Background 2 、 Integration of text information and graphic composite 3 、 Experimental results and discussion Outline

 Document Layout Analysis  Document Layout Understanding 1.1 Background DAR  To extract physical structure. Detection and labeling of the different zones (or blocks) as text body, illustrations, math symbols, and tables embedded in a document is called geometric layout analysis.

 Document Layout Analysis  Document Layout Understanding 1.1 Background DAR  To obtain logical structure. But text zones play different logical roles inside the document (titles, captions, footnotes, etc.) and this kind of semantic labeling is the scope of the logical layout analysis. Logical structure includes logic attributes, hierarchical relations and logical label association.

1.1 Background Image document based layout analysis and understanding Image doc

1.1 Background Digitized document based layout analysis and understanding Digitized doc

1.1 Background Mostly Solved problems for documents analysis in PDF formats: Text line and text block segmentation Table detection Formula detection Core detection Foot and head detection List recognition Paragraph recognition

1.1 Background Unsolved open problems for documents understanding in PDF formats:  Graphic recognition  Table recognition  Formula recognition  ToC  Reference detection  …

1 、 Background 2 、 Integration of text information and graphic composite 3 、 Experimental results and discussion Outline

2.1 Preprocessing Hierarchical For each document page, there are three files for description: A physical xml description of page elements and attributes, including text elements, image elements and path operations with its unique ID. A.png image with resolution of 300 dpi. This synthetic image is rendered according to the selected page. A labeled ground-truth file. It contains information for performance evaluation, such as bounding boxes and element IDs.

2.1 Preprocessing Hierarchical Multi-layer conception incorporating both structural representations and image based analysis is proposed for segmentation. The page images are divided into text layer and non-text layer. Text layer analysis. Clustering the text elements according to proximity of feature similarity. Non-text layer analysis. Connect component based graphic object segmentation.

2.1 Preprocessing Hierarchical Marginal pictorial decorations Decorative lines Photograp hic images Drawings integrating with text

2.2 Non-text layer analysis CC Connected Component detection is considered from visual perspective. Spatial arrangement of intensities described by image texture features is applied for graphic component segmentation. Gray level co-occurrence matrix. The value of indicates the frequency of value i co-occurs with value j in pre-defined spatial relationship.

2.2 Non-text layer analysis CC Local Texture Entropy. The entropy is highest when all entries in are equal. Morphological filtering. The morphological filter consisting conditional dilation is utilized to fill the holes.

2.2 Non-text layer analysis CC  Aiming at the reflowable reconstruction of PDF, the purpose of the graphical component segmentation lies in the rectangular bounding box of a holistic graphic composite which is mostly depicted by path operations in PDFs, rather than the fine edge boundaries of the detailed contents of graphic.  The outside bounding box of graphic object is then identified on the specific connected component. Up till now, the non-textual.png image I is segmented into N partitions R N. Each subregion R i, i=1,…,N is a connected component. In most cases, a whole graphic figure consists of multiple connected components. Further merging and splitting process are applied to group CCs into desired regions based on predefined criteria which is closely related with the inter text line space.

2.3 Text layer analysis Graph based

2.3 Text layer analysis Graph based Construct a graph G=, weight elements in a component should be similar so edges between two vertices in the same component should have relatively low weights. Elements in different components should be dissimilar, so edges between vertices in different components should have higher weights.

2.3 Text layer analysis Graph based is a component of G The internal difference: Difference between two components :

2.3 Text layer analysis Graph based Region comparison predicate: Maximum internal difference: where

1 、 Background 2 、 Integration of text information and graphic composite 3 、 Experimental results and discussion Outline

3.1 Experimental results Text

3.1 Experimental results Text

3.1 Experimental results Segmentation

3.2 Discussion Overlap