GripeLine by ED

Video Compression formats overview

At first glance, here lies a paradox:  S-Video signal with less than DV resolution, however, requires greater flow. The explanation is simple - in fact DV-encoding is slightly different from MJPEG. Thus, DV uses the 4:2:0 principle the idea that compared to the 4:2:2 save 25%. But most importantly, the algorithm of DV video format, as the development of MJPEG, uses a more flexible compression scheme based on adaptive selection of quantization tables. Compression ratio for different blocks, unlike MJPEG, changes in the image: for the very informative blocks (for example, at the edges of images) it increases, while for blocks with a large number of small parts is reduced relative to the average level of the image.  As a result, the same as achieved by reducing the amount of data by about 15%. At the same time, the main idea of DV compression format  is a constant standard video stream - 25 Mbps (Megabit Per Second) with fixed compression ratio - about 5:1.

Further reduction in the amount of data can be achieved by the transition to MPEG video format - a compression algorithm (MPEG - Motion Pictures Experts Group). It essentially focuses on the processing sequence of frames, and uses high redundancy of information in video frames, separated by a small time interval. Indeed, between adjacent video frames usually changes only a small part of the scene - for example, is smooth shift of a small object on the background of a fixed background. In this case, the full information about the scene can be stored selectively - for reference images.  For the rest enough to transmit only the difference information: the status of the object, the direction and magnitude of the displacement of the new elements of the background (the object as it moves). Moreover, these differences can be shaped not only by comparison with previous images, but also followed (as it is in them as you move an object called part of the background, previously hidden the object).  Thus, MPEG encoding essentially formed three types of frames: I (Intra), performing the role of supporting and preserving the full information about the structure of the video compression frame; P (Predictive), which carry information changes in the structure of an image compared to the previous frame (type I or P); B (Bi-directional), retaining only the most essential part of information on the differences from previous and subsequent images (only I or P). Schematic diagram of the subsequent compression of I-frames, as well as the difference P-and B-frames, similar to MJPEG, but, as in DV video compression format, with adaptive adjustment of quantization tables. It allows us to characterize DV video format as a special case of the MPEG sequence of I-frames with a given fixed flow (compression ratio).

Sequence of I-, P-, B-frames are combined into a fixed length and structure of the group shots - GOP (Group of Pictures). Each GOP must start with I and with a certain periodicity contains P frames. Its structure is described as M / N, where M - total number of frames in a group and N - the interval between P-frames. Thus, typical for Video-CD and DVD video format is as follows: IBBPBBPBBPBBPBB. Here, each B frame is restored on the surrounding P frame (in the beginning and end of the group - for I and P), and in turn, each P frame - on the previous P (or I) frame. At the same time I frames are self-contained and can be recovered independently from others, but are supporting for all P and B frames all the more groups.  Accordingly, I have the least amount of compression, the V - the greatest. Established that the size of a typical P-frame is 1 / 3 of I, and B - 1 / 8 part. As a result, MPEG sequence IPPP (GOP 4 / 1) provides 2-fold reduction in the required data stream (with the same quality) as compared with the sequence of only I frames, while the use of GOP 15 / 3 allows to reach 4-fold video compression.
 Included here for reference values of video streams that characterize the quality of movies recorded for Video-CD and DVD Video discs. Discussion of these recording formats is beyond the scope of this report, but will be considered later. Concluding the story of MPEG, it must be emphasized that this algorithm allows for variation, and many other encoding parameters, in particular, the spatial resolution. From this point of view distinguishes MPEG1, limiting the frame size of 352x288 and MPEG-2 video format, allowing different levels of resolutions (including 352x288), but as the main uses 720x576. Strictly speaking, MPEG1 is a limited version of MPEG2. However, given the resolution of 280 lines Video and involves the use of this limitation.

(c) All rights reserved.