Video Compression and Standard Codecs

The goal of video compression [9] is to remove the redundancy in the original source signal. This help in reducing the bandwidth necessary for transmitting the signal from point to another or allowing smaller space to store the video information. In multimedia real-time codecs, the goal of video compression is not only to minimize the bit rate of the given video signal, but also to guarantee some required levels of signal quality, to minimize the complexity of the used codec, and to reduce the overall encoding/decoding delay. However, it is not easy to satisfy these four requirements in the same time. Hence, a tradeoff between them is used to choose a compression scheme. Video compression is possible because of the following points: $\bullet$

• There is spatial redundancy among neighboring pixels in a given picture (frame),
• There is spectral redundancy among color images within frames,
• There is temporal redundancy among pixels or blocks of pixels in different frames,
• There is considerable irrelevant (from a perceptual viewpoint) information contained in video data.

A good compression scheme can exploit these four points to produce high compression rates and in the same time maintains a good quality level. Regarding the irrelevant information in video data, it is found that human eyes are not so sensitive to some non-redundant information. This is possible by employing lossless encoding techniques (like Huffman coding) and approximating the information, which is not important for the sake of human perception of the quality. For example, human eyes do not distinguish between two video sequences one of them encoded at 30 frames per second and the other at 60 frames per second. In addition, they are more sensitive to the variation of luminance than chrominance information. Thanks to the research in Human Visual System (HVS), there are nowadays many video encoders that make use of the above facts and provide reasonable quality and compression rates. In some situations (most of the current Internet video applications, for example), even by employing the previously mentioned encoding techniques, the output stream is not compressed enough to be fitted in the available bandwidth or the storage device. In this case the above techniques are mixed with another lossy encoding techniques to achieve certain compression ratio. In this case, the quality of the resulting stream is degraded with respect to the original video signal. This is good for some applications like the cheep Internet videoconferencing, video phones, or when some of the details of the video signal are not very important (video archiving for instance). In general, the more the quality degradation is accepted, the better the compression ratio is. The characteristics of the three types of video compression (lossless, lossy and hybrid) coding are given here. $\bullet$

• Lossless coding: It is a reversible process with the ability of perfect recovery of original data. Thus, video signal before lossless encoding is identical to that after decoding the encoded signal. Therefore, there is no quality degradation due to lossless coding. However, only low compression ratios can be obtained. An example of lossless coding is the ``Entropy coding''. The data is taken as a simple digital sequence. Some Entropy coding algorithms are run-length coding (RLC), Huffman coding, and Arithmetic coding.
• Lossy coding: It is an irreversible process in which the recovered data is degraded. Hence, the reconstructed video quality is sometimes degraded with respect to that of the original video. An example of lossy coding is the ``Source coding''.
• Hybrid coding: It combines both lossy and lossless coding. It is used by most multimedia systems. Examples of video codecs that use Hybrid coding are JPEG, H.263, and MPEG1/2/4.

We provide here the two kind of algorithms used to remove both spatial and temporal redundancies.