You are here
Basics of H.264/AVC Video Compression
We welcome you to the basics of H.264/AVC Video Compression training to Qt Quick for Mobile Development. The training will be given by Vinod Kumar Malamal Vadakital, on November 10 from 9.00 till 13.00 in Nokia office Hermia 4, Universum 172. Please don't forget to register yourself using the form in the bottom of this page.
The visual stream forms an important part of any multimedia content. However, the visual stream in its uncompressed form is made up of such large amount of data that it is beyond the capacity of any current transmission or storage device. Therefore, compression schemes are essential, and have been standardised. Most of these compression schemes use known human perceptual qualities along with efficient binary data coding schemes to reduce redundant information, remove irrelevant information, and perform compression. H.264/AVC, a joint video standard by the ITU and ISO, is among the popular video compressor that has been standardized. It has been shown to provide superior compression efficiency compared to any other earlier standards.
In this training, the participants are first introduced to the principles used to compress video data, and then a brief description on the workings of the various building blocks of a H.264/AVC video compressor. The training is divided into two sessions. In the first session, the theory is presented. In the second session the participants will use the reference JM coder and Matlab scripts to get a hands-on experience for video encoding and decoding. Selected features of the H.264/AVC coder will be evaluated using rate-distortion plots.
Training agenda
- Introduction to the concept and principles of video compression.
- Introduction to the H.264/AVC video codec blocks.
- Brief description of the workings of the different building blocks of the video codec.
- Understanding of the different compression concepts using practical experimentation.
Prerequisites
The JM reference software is available for free here. This software is written in C, and the latest version is JM18.0. Download the file jm18.0.zip and uncompress it. It contains, apart from the source code and documentation, Makefiles for a Linux platform, and MS visual C++ project files for a Windows platform. The participants must be able to download, compile, and produce an executable of this software. Matlab is also required for running the scripts. The participants should have some basic knowledge of Matlab and C.
