2D Object Detection and Recognition: Models, Algorithms and Networks

Citation

Rigelsford, J. (2003), "2D Object Detection and Recognition: Models, Algorithms and Networks", Sensor Review, Vol. 23 No. 1. https://doi.org/10.1108/sr.2003.08723aae.001

2D Object Detection and Recognition: Models, Algorithms and Networks

2D Object Detection and Recognition: Models, Algorithms and Networks

Yali AmitMIT Press2002 306 pp. ISBN 0-262-01194-8 £29.95 hardback

Keywords: Algorithms, Networks, Image processing

This book addresses two important aspects of computer vision, namely the detection and recognition of 2D objects. It presents a range of template models, techniques for their efficient implementation and how neural networks can be used to overcome variations in the object or the classifier.

The first chapter provides an introduction to image processing and presents topics including low-level image analysis and bottom-up segmentation, object detection with deformable template models, object recognition, and scene analysis. Chapter 2, Detection and Recognition: Overview of Models, discusses the Bayesian approach to detection, and overview of object detection models, and network implementations.

The following six chapters provide in-depth coverage of the detection algorithms. Chapters 3 and 4 present ID Models: Deformable Contours and Deformable Curves, respectively. The inside-outside model, an edge-based data model, joint estimation of the curve and the parameters, statistical models, and global optimisation of a tree structured prior, are amongst the topics discussed. 2D Models: Deformable Images, are addressed in chapter 5, while chapter 6 presents Sparse Models: Formulation, Training and Statistic Properties. The prior model and detecting pose are amongst the subjects discussed in chapter 7, the Detection of Sparse Models: Dynamic Programming. Chapter 8, Detection of Sparse Models: Counting, addresses detecting candidate centres, computing pose and instantiation parameters, density of candidate centres and false positives, and examples.

Chapter 9, Object Recognition, provides techniques for recognising isolated objects or shapes. It presents classification trees and object recognition with trees, relational arrangements, why multiple trees work, and experiments. Classification of chess pieces in grey-level images, and detecting and classifying characters are amongst the topics discussed in chapter 10, Scene Analysis: Merging Detection and Recognition, while Neural Network Implementations are presented in chapter 11. Topics discussed include basic network architecture, Hebbian learning and biological analogies.

The final chapter of the book provides a description of the software and datasets that are available on the associated web page. Most chapters conclude with bibliographical notes and a discussion.

“2D Object Detection and Recognition” is an informative text, which will appeal to computer scientists and researchers who are involved in modelling the functions of biological visual systems. Although the mathematical content of the book is not too advanced, the written style may make it less appealing to those who are new to image processing techniques.