With more and more images and videos being available in the compressed
format, researchers have started taking interest in the aspect of designing
algorithms for different image operations directly in their domains of representation.
This would not only avoid the inverse and forward transformation
steps and potentially make the computation faster but also would keep the
buffer requirement less as the storage required for processing the compressed
stream is less than that with its uncompressed representation.
This book attempts to comprehensively treat this topic of interest and
deals with the fundamentals and properties of various image transforms used
in image and video compression. Subsequently, their application in designing
image and video processing algorithms in the compressed domain are discussed.
To provide better understanding of the domain of research, different
image and video compression techniques are briefly covered in the first chapter.
In particular, discrete cosine transform (DCT) - based compression algorithms
(such as JPEG, MPEG, and H.264) and discrete wavelet transform (DWT) -
based JPEG2000 are discussed with more details. This is followed by discussion
on key properties of various transform spaces with special emphasis on
the block DCT space and the DWT.
Different types of image and video processing operations performed in the
compressed domain are discussed in subsequent chapters. This includes filtering,
enhancement and color restoration, image and video resizing, transcoding,
etc. In addition, different other applications in the compressed domain such
as video and image editing, digital watermarking and steganography, image
and video indexing, face detection and identification, etc., are briefly covered
in the last chapter.
This book is meant for readers who have gone through a first-level course
on digital image processing and are familiar with the basic concepts and tools
of image and video processing. However, at the introductory level, the details
of compression techniques and properties of the transform domain are not
always extensively covered. The first two chapters of this book discuss these
issues at considerable depth. For the sake of completeness, an effort has also
been made to develop concepts related to compressed domain processing from
the very basic level so that a first-time reader also does not have any problem
in understanding them.