The state-of-the art in computer vision: theory, applications, and programming

Whether you're a working engineer, developer, researcher, or student, this is your single authoritative source for today's key computer vision innovations. Gerard Medioni and Sing Bing Kang present advances in computer vision such as camera calibration, multi-view geometry, and face detection, and introduce important new topics such as vision for special effects and the tensor voting framework. They begin with the fundamentals, cover select applications in detail, and introduce two popular approaches to computer vision programming.

• Camera calibration using 3D objects, 2D planes, 1D lines, and self-calibration

• Extracting camera motion and scene structure from image sequences

• Robust regression for model fitting using M-estimators, RANSAC, and Hough transforms

• Image-based lighting for illuminating scenes and objects with real-world light images

• Content-based image retrieval, covering queries, representation, indexing, search, learning, and more

• Face detection, alignment, and recognition--with new solutions for key challenges

• Perceptual interfaces for integrating vision, speech, and haptic modalities

• Development with the Open Source Computer Vision Library (OpenCV)

• The new SAI framework and patterns for architecting computer vision applications

Amazon.com® Reader Reviews (Ranked by Helpfulness)

Average Amazon.com® Rating:  Based on 2 Ratings

Good supplemental text on computer vision - 2007-06-22
Reviewer Rating:
This is a good book as long as you realize it is intended to be a supplement to a good basic text on the topic of computer vision and not a textbook itself. As such, it is not a good source of algorithms. Instead, it takes a high level approach and discusses topics that computer vision textbooks don't have room to include. This is not to say that the book is page after page of narrative with no instruction on specific steps whatsoever. It is just not full of the matrices, transforms, and algorithms you would expect in a textbook on the subject. For example, when this book discusses using a neural network for some computer vision task, it is assumed you already know how to set up a neural network to solve a problem via MATLAB or some alternate method and that you know what kind of problems neural networks can aid in solving, where a computer vision textbook would likely go over the subject and science of neural networks separate from the specific concern of computer vision. The book is organized into three parts, covering various fundamentals, applications, and programming aspects of computer vision.

Part I, "Fundamentals In Computer Vision," consists of four chapters. Two of the chapters deal with the more conventional but still popular areas: camera calibration and multiview geometry. They deal with the most fundamental operations associated with vision. The chapter on robust estimation techniques will be very useful for researchers and practitioners of computer vision alike. There is also a chapter on a more recently developed tool, the tensor voting framework, that can be customized for a variety of problems.

Part II, "Applications In Computer Vision," describes a variety of interesting applications in computer vision, ranging from the more traditional fields of content-based image retrieval, face detection, and human tracking to more graphics-oriented areas of interest such as image-based lighting and visual effects. Chapter 6 describes how scenes and objects can be illuminated using images of light from the real world. While this operation, also known as image-based lighting, has its roots in computer graphics, it requires computer vision techniques to extract high dynamic range images and resample the captured light. Many of the special effects seen in movies rely on computer vision techniques to facilitate their production. Chapter 7 describes some vision techniques that have been used successfully in the movie industry. A natural extension to current text-based search engines would be image retrieval. Chapter 8 presents a survey on the theory and techniques for content-based image retrieval. The issues covered include interactive query formulation, image feature extraction, representation and indexing, search techniques, and learning based on feedback. Chapter 9 describes techniques for face detection, alignment, and recognition. They show how the difficult problems of changing head pose and different illumination can be handled. Chapter 10 describes perceptual interfaces, which involve the use of multiple perceptual modalities such as vision, speech, and haptic to enable human-machine interaction. The authors motivate the need for such a study and discuss issues related to vision-based interfaces.

One of the more overlooked areas in computer vision is the programming aspect of computer vision. While generic commercial packages can be used, there exist popular libraries or packages that are specifically geared for computer vision. Part III, "Programming For Computer Vision," describes two different approaches to facilitate programming for computer vision. This section gives very good detailed instructions on installing and using the open source packages described. The following is the book's detailed table of contents.

Chapter 1. Introduction
Part: I Fundamentals In Computer Vision
Chapter 2. Camera Calibration
Section 2.1. Introduction
Section 2.2. Notation and Problem Statement
Section 2.3. Camera Calibration with 3D Objects
Section 2.4. Camera Calibration with 2D Objects: Plane-Based Technique
Section 2.5. Solving Camera Calibration with 1D Objects
Section 2.6. Self-Calibration
Section 2.7. Conclusion
Section 2.8. Appendix: Estimating Homography Between the Model Plane and Its Image
Chapter 3. Multiple View Geometry
Section 3.1. Introduction
Section 3.2. Projective Geometry
Section 3.3. Tensor Calculus
Section 3.4. Modeling Cameras
Section 3.5. Multiple View Geometry
Section 3.6. Structure and Motion I
Section 3.7. Structure and Motion D
Section 3.8. Autocalibration
Section 3.9. Dense Depth Estimation
Section 3.10. Visual Modeling
Section 3.11. Conclusion
Chapter 4. Robust Techniques for Computer Vision
Section 4.1. Robustness in Visual Tasks
Section 4.2. Models and Estimation Problems
Section 4.3. Location Estimation
Section 4.4. Robust Regression
Section 4.5. Conclusion
Chapter 5. The Tensor Voting Framework
Section 5.1. Introduction
Section 5.2. Related Work
Section 5.3. Tensor Voting in 2D
Section 5.4. Tensor Voting in 3D
Section 5.5. Tensor Voting in ND
Section 5.6. Application to Computer Vision Problems
Section 5.7. Conclusion and Future Work

Part: II Applications In Computer Vision
Chapter 6. Image-Based Lighting
Section 6.1. Basic Image-Based Lighting
Section 6.2. Advanced Image-Based Lighting
Section 6.3. Image-Based Relighting
Section 6.4. Conclusion
Chapter 7. Computer Vision In Visual Effects
Section 7.1. Introduction
Section 7.2. Computer Vision Problems Unique to Film
Section 7.3. Feature Tracking
Section 7.4. Optical Flow
Section 7.5. Camera Tracking and Structure from Motion
Section 7.6. The Future
Chapter 8. Content-Based Image Retrieval: An Overview
Section 8.1. Overview of Chapter
Section 8.2. Image Domains
Section 8.3. Image Features
Section 8.4. Representation and Indexing
Section 8.5. Similarity and Search
Section 8.6. Interaction and Learning
Section 8.7. Conclusion
Chapter 9. Face Detection, Alignment, and Recognition
Section 9.1. Introduction
Section 9.2. Face Detection
Section 9.3. Face Alignment
Section 9.4. Face Recognition
Chapter 10. Perceptual Interfaces
Section 10.1. Introduction
Section 10.2. Perceptual Interfaces and HCI
Section 10.3. Multimodal Interfaces
Section 10.4. Vision-Based Interfaces
Section 10.5. Brain-Computer Interfaces
Section 10.6. Summary

Part: III Programming for Computer Vision
Chapter 11. Open Source Computer Vision Library (OPENCV)
Section 11.1. Overview
Section 11.2. Functional Groups: What's Good for What
Section 11.3. Pictorial Tour
Section 11.4. Programming Examples Using C/C++
Section 11.5. Other Interfaces
Section 11.6. Appendix A
Section 11.7. Appendix B
Chapter 12. Software Architecture for Computer Vision
Section 12.1. Introduction
Section 12.2. SAI: A Software Architecture Model
Section 12.3. MFSM: An Architectural Middleware
Section 12.4. Conclusion
Section 12.5. Acknowledgments

some topics are mature, others might be already obsolete - 2009-02-27
Reviewer Rating:
For machine vision researchers, the editors of the book compiled a good survey of the field in 2004. The book does not start from scratch, unlike Machine Vision by Davies. Instead, it dives straight into numerous topics, by assuming you are already versed in the basics.

The text has a combination of descriptions of the maths underlining the methods, and the showing of the results from applying the methods.

Some topics are by now fairly mature. Take image based lighting, where scenes are illuminated by one or more light sources. For realistic renderings, the methods described should give very good results.

Face detection, on the other hand, still has ways to go. The chapter on it talks about using Haar feature sets and other ideas. But the chapter may have been somewhat obsoleted by recent [2008] work that used another method that is orders of magnitude faster, though with roughly the same accuracy.

Browse Similar Topics

Top Level Categories:
Computer Science

Sub-Categories:
Computer Science > Theory Of Computation