Data compression#Video

Video compression is used to reduce the data rate of the digitized video signal to make it easier to store or downloading. Attainable compression rates are typically between 1:5 and 1:200.

The video compression has its origins in the still image compression, simpler method to compress the individual images of a video independently of each other and store the uncompressed sound from. The compression ratio achieved is approximately 1:5. More advanced methods use to encode similarities between the individual sub-images and also save the sound from compressed. The achievable compression rates are now above 1:100 at barely reduced quality.

The standardization of video coding method has become an international organizations about exciting process where the Moving Picture Experts Group ( MPEG) as the ITU are involved. Therefore, many identical processes have different names such as ITU H.264, MPEG- 4 Version 3 or MPEG -4 AVC, behind which is hidden the same codec.

Physiological basis of video compression

The compression algorithms are based on

• Redundancy of the video signal ( redundancy reduction ) and
• Deficiencies and physiological effects of human vision ( irrelevance ).

The redundancy reduction uses similarities between neighboring pixels in space and time and guess its value. Be coded only need the error. Reachable compression factors of 1:2 to 1:5. Irrelevanzkodierung discards the information that is not or hardly visible to the human observer. This further compression of typically 1:2 to 1:50 is possible, depending on process and the required quality.

Since the color resolution due to the anatomy of the eye is worse than the resolution of the difference of brightness, one can reduce the resolution of the color information without the differences could be strongly perceived. This is known as chroma subsampling. Many compression methods use this technique as a first step towards reduction.

A further property of the visual system, which can be utilized, the frequency dependency. You can take pictures, like the keyboard, even as a superposition of two-dimensional oscillations represent. Low frame rates are responsible for coarse image structures, high for fine details. Disturbances in the various frequency ranges are differentially perceived, which can be well illustrated by a simple test image.

This frequency dependence is used over the use of a suitable transformation in all video compression methods to the MPEG family.

Mathematical Foundations

The terms of redundancy reduction and irrelevancy derived from information theory and describe two different approaches for reducing the amount of data, also known as data compression, in the transmission of information. It draws on a model of mechanical transfer of information from a source to the sink. Applied to the specific case of video coding, the source corresponds to the sequence of video images as they occur in the original camera, the valley corresponds to the eye of the beholder.

Redundancy reduction

Redundancy reduction takes into account the characteristics of the source with the goal of reducing the amount of data to be transmitted. In the case of video encoding, statistical properties of the image signal, for example, correlation between adjacent pixels, temporally and spatially, utilized to produce compact code possible. The encoding is variable length codeword (VLC "variable length coding" ) are used. Rather than encode all symbols to be transmitted with constant codeword length, more frequent or probable symbols with shorter code words encoded as rarer symbols. Since no information is lost, this is called lossless coding.

Irrelevance

Irrelevance aims to omit that information during the transmission, which is not relevant to the sink. Specifically, this means in the case of video encoding, only a portion of the image data is transmitted. In this case, those distortions resulting be allowed to be in those perceived to the human observer as few disturbances. Since information is lost, this is called lossy coding.

Forward discrete cosine transform

In the forward discrete cosine transform ( FDCT ) the single video image (frame) is divided into 8x8 pixel blocks wide, and is judged according to their complexity. This step is necessary so that the codec "white", for which (complex) pixel blocks it needs a lot of memory and meet fewer bits for which (simple) blocks. This is the condition for irrelevance.

Motion correction

Another possibility for reducing the amount of data is the motion compensation ( English motion compensation ): Only the differences from the previous image stored. It searches for the pixel blocks that have changed since the last frame. For this, a motion vector is stored, the non-moving are simply taken from the last frame.