Video compression formats
Before building a video surveillance system, a technician will have to solve a number of critical problems and tasks. Along with the choice of cameras, servers, and software, it is necessary to select the optimal compression format for the system’s video transmission. The main formats for video surveillance systems are MJPEG, MPEG-4 and H.264 (MPEG-4 part 10). Controversy “Which format is better?” Similar to the disputes “what is the future for: analog or ip?”, have been going on for several years, but experts still cannot reach a consensus. In this article we will talk about the characteristics of the formats, the parameters that must be taken into account when choosing, and the technologies used to analyze video streams.
Compress video
The videos from the IP surveillance cameras reach the server in compressed form. Compression generally involves removing features that are almost invisible to the human eye, although compression is sometimes done without loss of information. Compressed video stream requires less network bandwidth and less free hard disk space. To view or analyze the video, the resulting stream must be unzipped; apply the reverse conversion algorithm to compression. The combination of compression and decompression algorithms is called a video codec. Video codecs of different standards are not compatible with each other; video information compressed with one codec, as a rule, cannot be decompressed with another.
There are several types of compression.
Lossless compression allows you to obtain an image after decoding that does not differ from the original frame.
Lossy compression loses information after decoding.
Several lossy compression implementations are possible:
lossy percentage compression – loss is so insignificant to the human eye that the frame before and after decoding is virtually indistinguishable to the operator;
lossy compression – the differences between the frames before and after decoding are noticeable, but still not very pronounced, all the information necessary to analyze the events is saved;
Lossy Compression: Low-quality streaming video compression, causing artifacts (noticeable video distortion) during decoding. The appearance of artifacts leads to a decrease in image clarity, the appearance of fields of the same color (when different color tones are combined in one), the appearance of image blockages (pixelation, graininess). The presence of artifacts leads to false results of the analysis of the video stream by the system software.
In the compression process, to reduce the size of the video stream, the amount of color tones in the image are reduced, color resolutions are lowered, and small details in the image that are invisible to the human eye are removed; predict changes based on data already received; remove duplicate pixel values.
Video compression formats
There are many compression formats, MJPEG, MPEG-4 and H.264 are the most popular in video surveillance.
MJPEG format
For the MJPEG compression format, a video sequence is a sequence of still images: JPEG images. Compression occurs individually for each frame (intraframe). We get total independence from individual images. When playing a video file, the image quality is still good: from the MJPEG format you can always get frames with a clear image of the events that take place, it does not require high processor performance, but significantly loads the network and requires a large amount of disk space. This format is characterized by image blocking artifacts, fields of the same color. Camera data is lossy, so it’s impossible to say there is no distortion. Another thing is that if the camera is correctly adjusted, the human eye in JPEG hardly notices distortion.
MPEG-4 and H.264 formats
For MPEG-4 and H.264 formats, compression is performed both within a frame and for a series of frames (between frames). H.264 video (MPEG-4 optimized or MPEG-4part 10) is not a sequence of individual images, but a chain of related data – video streaming. The advantages of this format are that not all frames are saved, but only the reference image and its subsequent changes.
When a significant part of the image remains unchanged, the resulting video size is much smaller than for MJPEG. In case the MJPEG format can send a set of images of 200 KB each, the H.264 format will send a reference image of 200 KB and its subsequent changes, which are much smaller.
