Customer Reviews

Digital Video Processing

5 star:		(2)
4 star:		(0)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

This book is really good. Starting from the very first chapter, this book takes you into a very interesting world of digital video. It discusses MPEG video and the technology behind in some detail. This is for the people who want to know how digital video is created and the details about MPEG-2 video. The best part of the book is the Motion Estimation and Compensation portion...... Very high level of mathematical stuff is used in the book and hence will require some good reference books too....

Although published in 1995, this is still the book I come back to time and time again whenever I have a question about processing digital video. Unlike other titles, it is neither knee-deep in terse useless theory nor is it a book about which connector goes where in digital video. You might call it the Gonzalez & Woods of digital video processing, in honor of those authors' fabulous book on digital image processing. The book is organized into six parts:

Part 1- "Representation of Digital Video" consists of chapters 1 through 4. Chapter one is an introduction to both analog and digital video. Chapter 2 models the formation of time-varying images as perspective or orthographic projections of 3-D scenes with moving objects. Chapter 3 addresses the time-space sampling of 3-D lattices, which is what a progression of video frames actually is. Conversion between sampling structures without using motion information is tackled in chapter 4. This is the task of converting video from one format to another.

Part 2 - "Two Dimensional Motion Estimation" consists of chapters 5 through 8. Motion estimation is one of the most effective ways to utilize time redundancy in video frames, and this section discusses four ways to do this - optical flow, block-based methods, pel-recursive methods, and Bayesian methods.

Part 3 - "Three Dimensional Motion Estimation and Segmentation" consists of chapters 9 through 12. The models in this section can be used for either 3-D image analysis or for improved 2-D motion estimation. The two-view estimation techniques described in chapters 9 through 11 are based on optical flow and are very sensitive to small inaccuracies that are present. Thus the method of chapter 12 uses stereo pairs and motion tracking over long stereo sequences for more robust results.

Part 4 - "Video Filtering" consists of chapters 13 through 17. Video filtering differs from image filtering due to the motion information involved. Thus motion-compensated filtering is discussed in chapter 13. Motion-compensated filtering to reduce the graininess found in most "freeze-frame" images is discussed in chapter 14. Restoration of blurred video frames is the subject of chapter 15. Chapter 16 involves conversion between video standards and discusses frame/field up/down conversion and interlacing/deinterlacing techniques. Chapter 17 discusses a comprehensive model for low-resolution video acquisition and a framework for superresolution of video.

Parts 5 and 6, the remainder of the book, discuss still image compression and video compression respectively. These sections are detailed, but more recent developments in compression such as MPEG-4 and wavelets are omitted due to the age of the book.

The reader should already be familiar with image processing techniques as well as calculus, probability, matrix algebra, and some differential equations to get the most from this math-intensive book. However, the book is very detailed to the point that it is straight-forward to program solutions based on its contents. There are exercises at the conclusion of each chapter, but no solutions, since most of the exercises consist of essay-type questions or derivations. This seems odd since the book is very practical in its exposition, one would assume that the exercises would be more practical in scope. I highly recommend this book for anyone interested in digital video processing.