This book is divided into three parts.Part Irepresents polygonal approximation of closed digital curves. Part II deals with scale-space analysis with its application to corner detection.PartIlI demonstrates series of case studies on structural pattern clas- sification and 2-D object recognition.
Approximation of a closed curve by piece straight line segments is known as po- lygonal approximation.Any curve can be approximated by apolygon with any desired degree of accuracy.
The aim of this book is to introduce to the beginner the representation of two-di- mensional objects in terms of meaningful information that may be used for recognition of structural patterns,2-D objects,scene and/or occluded scene.The book is basically an outcome of the research performed by the vision research group under Professor Kumar S.Ray at the Electronics and Communication Sciences Unit,Indian Statistical Institute, Kolkata.
This book has tremendous importance in the area of structural pattern classifica- tion, object recognition, computer vision, robot vision, medical computing, compu- tational geometry, and bioinformatics systems. The systematic development of the concept of polygonal approximation of digital curves and its scale-space analysis are useful and attractive.Development for different algorithms of polygonal approxima- tion and scale-space analysis and several experimental results with comparative study for measuring the performance of the algorithms are extremely useful for theoreti- cal and application-oriented works in the above-mentioned areas.Polygonal approxi- mation of digital curves has been cultivated by many researches, but its scale-space analysis has hardly been discussed.Hence,a systematic development of polygonal ap- proximation and scale-space analysis of digital curve is very essential for future work. Also there is no comprehensive textbook on polygonal approximation and scale-space analysis of digital curves.Hence this book is very useful from theoretical and practical points of view.It simply fills the gap.Finally this book describes a series of applica- tions of the methodology of polygonal approximation in the specific areas ofstructural pattern classification and recognition of 2-D scene and/ or 2-Doccluded scene.
This book is aimed at academics, graduate, postgraduate students,and research professionals.It is a book that can be used in coursework on image processing,pattern recognition,and computer vision at graduate and post-graduate levels in computer science/electrical engineering/electronics engineering.
We have not tried to trace the full history of the subject treated; this is beyond our scope. However, we have assigned credits to the sources that are as readable as pos- sible for one knowing what is written here.A good systematic reference is covered in the list of references of the book.
We would like to thank Mandrita Mondal for preparing the materials of the book.— Kumar S. Ray, PhD,and Bimal Kumar Ray, PhD

List of Contributors vii

List of Abbreviations xi

Preface xii

1.Polygonal Approximation. 1

2.A Split and Merge Technique 11

3.A Sequential One-pass Method 21

4.4.Another Sequential One-pass Method3 1

5.A Data-driven Method 45

6.Another Data-driven Method 57

7.A Two-pass Sequential Method 67

8.Polygonal Approximation Using Reverse Engineering on Bresenham's Line Drawing Technique 95

9.Polygonal Approximation as Angle Detection 103

10. Polygonal Approximation as Angle Detection Using Asymmetric Region of Support 113

11.Scale Space Analysis with Application to Corner Detection 125

12. Scale Space Analysis and Corner Detection on Chain Coded Curves 129

13. Scale Space Analysis and Corner Detection Using Iterative Gaussian Smoothing with Constant Window Size 143

14.Corner Detection Using Bessel Function as Smoothing Kernel 175

15. Adaptive Smoothing Using Convolution with Gaussian Kernel..191

16. Application of Polygonal Approximation for Pattern Classifcation and Object Recognition 199

17.Polygonal Dissimilarity and Scale Preserving Smoothing 203

18.Matching Polygon Fragments 221

19. Polygonal Approximation to Recognize and Locate Partially Occluded Objects Hypothesis Generation and Verification Paradigm 235

20. Object Recognition with Belief Revision: Hypothesis Generation and Belief Revision Paradigm 259

21. Neuro-fuzzy Reasoning for Occluded Object Recognition: A Learning Paradigm through Neuro-fuzzy Concept 311

22.Conclusion 341

Index 369

中科院文献情报中心