Information Technology Reference
In-Depth Information
edges. In case of H.264, a de-blocking filter is applied at the decoder to resolve
discontinuities.
The wavelet transform has been widely applied to still images as well. It is an
observation that the Discrete Wavelet Transform (DWT) outperforms a DCT by
an average 1dB [9]. Wavelet transform was selected in JPEG2000 and MPEG-4
due to its better energy clustering properties and it is also frame based [2].
Different error resilient algorithms have also been proposed that are based on
partitioning of Wavelet Transform Coefficients [15].
2 Advantages of Video Event Analysis in Wavelet
Video Data coded using wavelet transform offer several new avenues to video
event analysis.
Video event analysis is generally carried out on raw (un-coded) video data. As
video data is large it is usually available in coded form and requires decoding first.
As wavelet transformed image data also contain a low resolution image it offers
avenues for direct video analysis on this image, hence offering speed advantages.
Similar analysis can be contemplated for the intra coded wavelet video.
Wavelet transform enables the extraction of multiple resolution images from a
single coded data stream. This makes it easy to extract a lower resolution or lower
quality video from the compressed code-stream. This is specifically true for
coding paradigm where the wavelet transform is applied to the complete images
rather than on image blocks. This offers avenue for performing multi-resolution
image analysis leading to identification of multi-resolution events in video. We
will therefore, discuss how scalable video can be obtained in case of Dirac after
discussing the architecture of Dirac in detail.
3 BBC Dirac Video Codec
Dirac is an experimental Open Source video codec initially developed by BBC
Research. It is a royalty-free video codec, named in the honor of the British
Scientist Paul Dirac. The aim of the Dirac was to build up a high performance
video compression codec with a much simpler and modular design both
conceptually and in implementation. Dirac was required to be able to encode and
decode the videos in real-time, however, it is too slow for real-time conferencing
in its current state. Dirac documentation including high level algorithmic
description, doxygen-based comments on the reference code, specification of
Dirac bit stream and decoding are provided with Dirac video codec. Dirac is also a
wavelet based codec. This topic chapter discusses the architecture, features and
algorithmic description of Dirac in detail. It also highlights different directions of
ongoing and future research. In [10], different approaches for optimizing Dirac are
also described.
Search WWH ::




Custom Search