Application of AVS in video surveillance system

With the rapid development of computer, video coding and network transmission technologies, video surveillance technology has achieved a huge leap. Video coding technology has attracted everyone's attention as a key technology for remote video surveillance. At present, the image compression standards H.261 and MPEG-1 that are mainly used in video surveillance systems have limitations in application and poor user interaction. The MPEG-2, MPEG-4, H.263, H.264 and other video coding standards produced in recent years provide a higher data compression ratio, especially H.264 with its high quality, low bit rate, and strong adaptability The characteristics of are more and more widely used in video surveillance systems.

Because the above standards have certain problems in terms of patent pool fees, intellectual property rights, and complexity of technology implementation, they restrict their development. Whether we can establish a video coding standard with independent intellectual property rights in China and the performance is the same or better than other standards has become an important issue facing us. In this case, AVS (Audio Video Coding Technology Standard) came into our sight.

1 AVS video coding standard

(1) AVS main technical improvements

AVS uses many advanced technologies to ensure its performance, which is a new standard with more advantages. The AVS standard part 2 (AVS1-P2 hereinafter referred to as AVS video standard) completed at the end of 2003 is mainly for high-definition, high-quality digital TV broadcasting, digital storage media and other related applications. It has four characteristics: high performance, coding efficiency is more than 2 times higher than MPEG-2, and H.264 coding efficiency is equivalent; algorithm complexity is lower than H.264; software and hardware implementation costs are lower than H.264; patent The authorization model is simple and the cost is significantly lower than similar standards.

The AVS video standard uses a technical framework similar to H.264 (see Figure 1), including technical modules such as transformation, quantization, entropy coding, intra prediction, inter prediction, and loop filtering.

figure 1

The main advanced technologies to realize the AVS video standard are as follows.

Average information coding

First, AVS adopts k (k = 0,1,2,3) order Exp-Golomb (exponential Columbus) coding, CBP (macroblock coding template), macroblock mode and motion vectors adopt 0th order exponential Golomb coding, The quantization coefficient uses all four kinds of exponential Golomb codes, and uses 2D-VLC to encode it. Due to the adjustment of the Exp-Golomp encoding tables, the AVS decoder does not need to store these encoding tables. Syntax elements can be decoded using simple analysis with selectable lookup tables. Although the 19 mapping tables defined by AVS only occupy less than 2Kbyte of space, they can adapt well to different allocations and have high coding capabilities.

? Conversion and quantization

Unlike H.264 and MPEG-2, AVS uses 8 & TImes; 8 integer conversion. In order to reduce the rounding errors in dequantization and inverse conversion, AVS also sets up a special program, and various operations can be completed within 16 bits.

Intra prediction

The AVS video standard uses intra prediction technology to improve the performance of intra-coded macroblocks. Compared with H.264, AVS defines 5 modes for 8 & TImes; 8 luma blocks and 4 modes for 8 & TImes; 8 chroma blocks.

Reference screen

In previous video coding standards (such as MPEG-2), bidirectionally predictive coded pictures (B pictures) are usually based on the previous picture and / or the next picture. Although the predictive coded picture (P picture) uses only the previous picture to predict the current picture, the actual reference buffer space in the decoder is equivalent to twice the size of the picture. AVS completely uses the reference buffer for P buffer encoding. P pictures can be referenced by two adjacent I / P pictures. Therefore, while improving encoding efficiency, the reference buffer space is still the same as MPEG-2 .

B picture symmetry mode

AVS replaces the interpolation mode of the existing coding standard with a symmetric mode, only forward motion vectors are encoded, and reverse motion vectors are derived from the mutual relationship between forward and reverse. Therefore, at least one direction of motion vectors needs to be encoded in the B macroblock of AVS.

Weighted prediction

The weighted prediction function of AVS can greatly improve the coding efficiency, especially when the scene changes and the lighting changes. Weighted prediction uses a simple linear mode, so the parameters can be encoded in the predicted image header. However, each macroblock can freely choose whether to use weighted prediction.

Deblocking filter

Block-based video coding often produces block artifacts, which becomes more apparent at low bit rates. To solve this problem, AVS defines an adaptive loop deblocking filter to improve decoded video quality. In addition to picture boundaries or slice boundaries, the filtering function can also be used for the boundaries of luma and chroma blocks. The strength of filtering depends on the type of macroblock, quantization order, motion vector, and the difference between blocks.

Interlaced coding

When inputting the interlaced sequence, a picture can be encoded with one frame or two fields (top field and bottom field). Only picture layer adaptation between frames and fields can be used in the current version. When two-field coding is used, the previous field is predicted by the previous decoding field, and the next field is jointly predicted by the previous field and the previous decoding field. The two scenes share an image header, but they should belong to different films.

(2) Comparison of AVS and H.264

Video information is usually massive information, which requires a lot of storage space and transmission bandwidth. The key to the study of video standards is to optimize algorithms, eliminate redundant video data, and compress video data to the maximum extent without distorting the visual effect of the image.

Compared with the current best H.264 standard, under similar image quality, the AVS decoding complexity is equivalent to 30% of H.264, and the AVS encoding complexity is equivalent to 70% of H.264. The lower the complexity, the less system resources are occupied, and the system efficiency is increased accordingly. The complexity analysis of specific technical modules is shown in Table 1.

2 Application of AVS in video surveillance system

(1) Structure of video surveillance system

Generally, the video surveillance system can be divided into three parts: front-end equipment, communication equipment and back-end equipment. Its structure is shown in Figure 2.

figure 2

Front-end equipment mainly includes video servers and other related equipment. The gimbal controller can control the camera to rotate and focus, and has functions such as video switching, receiving alarm, and receiving temperature measurement. The video server is responsible for digitizing the video and compressing the image through video encoding, and can adjust the PTZ controller through the received front-end control signal.

Back-end equipment mainly includes video surveillance servers and surveillance hosts. The video monitoring server receives the compressed video sent from the front-end video server and other alarm, temperature and other information, and then forwards it to the corresponding monitoring host. The monitoring host sends control commands based on the obtained monitoring information.

(2) Application of AVS in video surveillance system

The video digitization and video encoding compression functions are implemented in the front-end video server. We can use the AVS standard to encode video information and decode it in the back-end device. The advantages of this are as follows.

Compression ratio and image quality

In the video surveillance transmission system, in order to occupy the transmission bandwidth together with other data, in order to ensure the quality of the video image, the bandwidth occupied by the transmission video needs to be as small as possible, which requires efficient video coding and decoding and high compression rate.

The AVS standard can provide continuous and smooth high quality, and under the same picture quality, the compression ratio of AVS and H.264 is similar, which saves 64% of the code stream on average compared to the previous generation encoding standard MPEG-2, and saves on average than MPEG-4 39% of the transmission code stream, and through the improvement of traditional intra prediction, inter prediction, transform coding and entropy coding algorithms, the AVS coding efficiency and image quality are further improved on the basis of previous standards.

Network adaptability

In the video surveillance transmission system, depending on the network, the available transmission bandwidth is also different, so how does AVS video adapt to different bandwidth networks for transmission?

AVS stipulates different grades and levels. The grades and levels impose various restrictions on the bitstream, and also specify the coding and decoding capabilities required for a particular bitstream coding and decoding. The grade is a subset of the specified grammar, semantics and algorithms. Coders and decoders that conform to a certain grade must fully support the subset defined by the grade; the level is the restriction on the syntax elements and syntax element parameter values ​​under a certain grade Sets, at a given level, different levels often mean different encoding and decoding capabilities and memory capacity. We can adopt different grades and levels according to different network transmission capabilities and application quality requirements.

Sensitivity and robustness of motion detection

The sensitivity and robustness of motion detection are the key indicators of video surveillance systems and the basis of image tracking and real-time alarms. In traditional motion detection methods, a single criterion (signal energy threshold control) is generally used to determine the occurrence of motion. Now we can use AVS motion compensation technology to detect motion in video surveillance. Such as: AVS standard motion prediction has 1/4 pixel accuracy, the calculation of motion vectors is more refined, and the motion information in the image is better portrayed; AVS standard uses weighted motion compensation technology, which can remove local light and dark changes and local light Or the similarity and relevance brought by the slight jitter of the overall picture.

Independent standardization

AVS is an audio and video coding technology standard independently developed by China that is suitable for a wide range of digital video use. It is an open standard based on independent technology and some open technologies. It properly solves the problem of patent licensing and takes into account the contradiction between performance and complexity . China's increasingly strong industrial strength and market also provide a good ground for its development.

3 Conclusion

AVS, as an audio and video coding standard independently developed by our country, has obvious advantages over other source standards, especially the characteristics of high coding efficiency, relatively low implementation complexity, and low patent costs, which will effectively promote the development of video surveillance technology. .

IC Sockets

Ic Sockets,Ic Sockets Connector,Ic Straight Pin Socket Connector,Round Pin Ic Sockets Connector

Dongguan ZhiChuangXing Electronics Co., LTD , https://www.zcxelectronics.com

Posted on