The book focuses on one of the key issues in document image processing – graphical symbol recognition, which is a sub-field of the larger research domain of pattern recognition. It covers several approaches: statistical, structural and syntactic, and discusses their merits and demerits considering the context. Through comprehensive experiments, it also explores whether these approaches can be combined.
The book presents research problems, state-of-the-art methods that convey basic steps as well as prominent techniques, evaluation metrics and protocols, and research standpoints/directions that are associated with it. However, it is not limited to straightforward isolated graphics (visual patterns) recognition; it also addresses complex and composite graphical symbols recognition, which is motivated by real-world industrial problems.
Author(s): K.C. Santosh
Publisher: Springer
Year: 2018
Language: English
Pages: 184
Foreword......Page 6
Preface......Page 8
References......Page 9
Acknowledgements......Page 10
Contents......Page 11
About the Author......Page 14
Acronyms......Page 15
1.1.1 What is Document Imaging?......Page 17
1.1.2 Basics to DIA and Challenges......Page 18
1.2 Graphics Processing......Page 24
1.3 Summary......Page 26
References......Page 28
2.1 Graphical Symbols......Page 32
2.2 Basics to Graphics Recognition......Page 33
2.3 Contests and Real-World Challenges in Graphics Recognition......Page 34
2.4 Graphical Symbol Recognition, Retrieval, and Spotting......Page 39
2.5 Research Stand Points: A Quick Overview......Page 40
References......Page 44
3.1.1 Data Acquisition and Preprocessing......Page 50
3.1.2 Data Representation and Recognition......Page 51
3.2 Validation......Page 53
3.2.2 Evaluation Metric......Page 54
3.2.4 Retrieval......Page 56
3.3 Summary......Page 61
References......Page 62
4.1 Statistical Pattern Recognition: A Quick Review......Page 67
4.1.1 Contour-Based Shape Analysis......Page 68
4.1.2 Region-Based Shape Analysis......Page 69
4.2 Graphics Recognition......Page 70
4.3 Experiments......Page 77
4.3.1 DTW-Radon: How Does It Work?......Page 78
4.3.2 Results and Comparison......Page 83
References......Page 90
5.1 Context......Page 95
5.2 Visual Primitives......Page 96
5.3 Spatial Relations......Page 98
5.3.1 Types of Spatial Relations......Page 101
5.3.2 Can We Quantify Spatial Relations?......Page 105
5.4 Structural Approaches for Graphics Recognition......Page 106
5.5 Spatial Relations on Graphics Recognition......Page 110
5.6 Can We Take Complex and Composite Graphical Symbols into Account?......Page 112
5.6.1 Symbol Recognition Using Spatial Relations......Page 113
5.6.2 Extension: Symbol Spotting......Page 117
5.7 Summary......Page 122
References......Page 125
6.1 Context......Page 134
6.2 Hybrid Approaches for Graphics Recognition......Page 136
6.3 Integrating Shape with Spatial Relations for Graphics Recognition......Page 137
6.4.1 Graph via Visual Primitives......Page 139
6.4.2 Shape-Based Thick Pattern Description in Arg via Clustering......Page 141
6.4.3 Cluster Verification and Validation......Page 144
6.5.1 Graphical Symbol Recognition......Page 148
6.5.2 Results......Page 150
References......Page 154
7.1 Syntactic Approaches-Based Graphical Symbol Recognition......Page 157
7.2.1 Basics to ILP......Page 159
7.2.2 How Does ILP Work?......Page 160
7.2.3 ILP for Character/Text Recognition......Page 161
7.3.1 Overview......Page 163
7.3.2 Graphical Symbol Representation......Page 164
7.3.3 Graphical Symbol Recognition......Page 167
7.4 Summary......Page 169
References......Page 170
8.1 Summary State-of-the-Art Works and Extensions......Page 174
References......Page 179
Index......Page 181
