In digital surveillance industry, video surveillance is important. It can record necessary video into storages. Because time is irreversible, we can only understand what happened in the past by the recorded images. Therefore, the key points of video surveillance are to provide correct stable real-time video recording and storage. High efficient video compression algorithm plays an important role. ...
In digital surveillance industry, video surveillance is important. It can record necessary video into storages. Because time is irreversible, we can only understand what happened in the past by the recorded images. Therefore, the key points of video surveillance are to provide correct stable real-time video recording and storage. High efficient video compression algorithm plays an important role.
H.264 becomes more and more important in nowadays video compression algorithm researches. With distinguished high efficient and video quality, it is generally acknowledged as best video compression algorithm nowadays. The compression rate for H.264 is about 9 times than Motion JPEG. Comparing with MPEG2 utilized in DVD, compression rate for H.264 is about 3 times than MPEG2. In other words, in the same video quality and resolution requirement, the file size becomes one third if H.264 instead of MPEG2 is utilized. With H.264, the storage of traditional DVD may contain more than three movies in the same DVD disk. Even comparing to MPEG4, H.264 is about 1.5 to 2 times better than MPEG4. With such a good efficiency, many cutting edge video systems use H.264 as their video compression standard, such as Bluray disk format and High-definition digital TV broadcast systems.
However, the trade-off of high compression rate of H.264 is huge computing requirement. By the estimation by some researchers, the computing requirement for H.264 compression is about ten times than MPEG-2. Therefore, maybe some personal computers can real-time decode H.264 video, but even the newest personal computers are still hard to achieve real-time H.264 encoding. Because what video surveillance needs is encoding, it becomes important to provide such high computing resource for H.264 encoding.
It is worth noting is that some tricky methods may be found in some so-called H.264 solutions. They mention that they adopt H.264 algorithm but skip some important steps in algorithm to avoid huge computing requirement, and then make their files cannot be decoded in standard H.264 codec on PC. This kind of solutions may be called H.264-like solution, but not the true H.264 solutions. Without compliance to standard H.264 algorithm, it may cause problems in CMS system integration.
Special Features of Video Compression in Surveillance System
Besides to the efficiency of video compression algorithm, we need to consider the special features in surveillance systems. Because video surveillance is widespread utilized in various applications, the image sources and requirements are much different. The video source may come from the main hall in an airport, and the image is full of crowded people and well-lighted. The source may come from some industrial factory, such as an oil field, and the image is dark and seldom changing. The source may come from vehicles, and the image is shaking and fuzzy. The video may also be used to avoid fraud in a casino, and the image must show all detail in order to see through the tricksterˇs tricks. The key points in different applications for video surveillance are diverse. Therefore, the solution for video surveillance system must remain flexible.
On the other hand, itˇs noticeable that surveillance industry still lacks of system standard. For other industries related to video compression, most of them have the system standards, such as optical disk (DVD, Bluray) and TV broadcast (includes broadcast via satellite, terrestrial wave, cable, IP network and mobile phone). In the system standard, not only the video compression but most of system components are standardized, such as data transmission, broadcast, storage and control signal, etc. In a well-defined system standard, the interoperation among components wonˇt be a problem. However, in current surveillance industry, the system is not standardized yet. Hence the flexibility is very important to fit different integration requirements.
Solutions to Realize H.264 Codec
Realizing H.264 codec requires enormous computing resources, especially the real-time applications. It is important to offer such a great computing resources to fulfill the requirements. First, the traditional "Soft Compress Card" method becomes unfeasible in H.264. This method makes use of PC software for low-cost compression and works well on Motion JPEG compression. However, because H.264 encoding is far behind the capability of personal computer, this method is no more a solution in H.264 era.
To realize H.264 codec, there are several methods. The first method makes use of high performance VLIW DSP. VLIW (Very Long Instruction Word) is a technique to boost DSP performance by instruction design, and make DSP can provide higher computing resource than CPU and traditional DSP. Nevertheless, due to instruction design change, it increases the difficulty for software programmers for software writing. Another issue is power consumption. Some VLIW DSPs have side effect of high power consumption, and make the IC temperature too high.
Another method is ASIC solution, which implements H.264 algorithm by hardware circuit on IC. For system developers, if the ASIC IC fit their need, this method is good for them because they donˇt need to handle algorithm inside IC. Because TV broadcast and optical disk industries have well-defined system standard, the system developers in TV and disk industries often choose ASIC IC. However, the ASIC is pre-programmed and lacks of flexibility. It cannot be modified to fit some special need. Besides, if the surveillance system will be scaled or upgraded in the future, the ASIC solution may not be able for future need.
The third is hybrid method, and it combines the advantages of both VLIW DSP and ASIC. For H.264, the kernel functions of algorithm are fixed and seldom changed, so they can be hardwired as ASIC blocks. Then, the VLIW DSP can be utilized to thread up whole algorithm with ASIC kernel functions. This method can give consideration to both efficiency and flexibility. Any modification can be achieved easily by software programming on VLIW DSP, while the hard computing requirement can be fulfilled by ASIC blocks to reduce power consumption and then lower the IC temperature.
Sunplus Coreˇs H.264 codec solution is hybrid method based on proprietary VLIW DSP and ASIC technologies. It can achieve standard H.264 standard codec functions in low power consumption, and offer flexibility for difference system needs. Due to the programmability of VLIW DSP, our solution can even offer computing of intelligent surveillance algorithm.