GripeLine by ED |
||
Video compressionIt is known that the television signal is a combination of the luminance signal Y and two chroma signals U and V. Variations in their values allow 256 levels (from 0 to 255 for Y, and -128 to 127 for U / V), which in binary terms, corresponding to 8 bits (bit) or 1 byte (Byte). Theoretically, every element of the frame has its own value YUV, i.e. requires 3 bytes. This representation, while the brightness, and chrominance signals have an equal number of independent values, usually referred to as 4:4:4. However, it was found that the human visual system is less sensitive to the color space changes than to luminance. And without any visible loss of quality the number of color samples in each row can be reduced by half. That such a representation, denoted as 4:2:2, was taken in a professional television. In this case, U-V-matrices are reduced to 360x576, and to transfer the full value of the television signal in each sample frame enough 2 bytes (alternating through independent count values of U and V). But for purposes of consumer video was appropriate in halve and vertical color resolution, i.e. go to the 4:2:0 coding. This reduces the color matrix to 360x288 and reduced number of bytes per sample - up to 1.5. This representation was included in the DV-format digital cameras format. Thus, taking into account the television frame rate of 25 Hz, we conclude that one second of digital video in the 4:2:2 representation requires 25x2x720x576 = 20736000 bytes, i.e. data stream is 21 MBps (MegaByte Per Second), but the 4:2:0 representation reduces the stream by 25% - up to 16 MBps. Record of such digital DV streams is technically feasible, but complicated, expensive and inefficient from the standpoint of subsequent processing. Real opportunities for practice require a considerable decrease in flow, i.e. forced to use different types of compression. There are many algorithms for performing compression without loss of information, but even the most effective ones for typical images do not provide the video compression ratio of more than 2 times. Among the video compression formats with data loss by one of the best known is MJPEG (Motion-JPEG). He came out of digital photography, where under the name JPEG was designed for compressing individual frames (JPEG - is an abbreviation of the name of the confirming international association of Joint Photographic Experts Group). Prefix Motion merely reflects his application for a sequence of frames, although each of them is processed independently. In this algorithm, the frame is divided into blocks of 16x16, each of them using the inverse discrete Fourier transform (IDFT) is carried the frequency domain. As a result, the distribution of luminance and chrominance (4:2:2, the representation) is transformed into the corresponding frequency coefficients, which are then subjected to quantization (rounding values given by the interval). In itself, the IDFT is reversible - doesn't lead to data loss, but here the quantization of the coefficients causes deviation of the image. Quantization operation is performed with varying intervals - the most accurate low-frequency information is transmitted, since the corresponding distortion of the image visually small. At the same time, many high- factors responsible for the "fine" details of the image after it takes zero values.
|
||
(c) Gripe2ed.com. All rights reserved. |
||