The full name of the H.264 standard is "H.264 / MPEG-4 part 10", which is a new standard formulated by the Joint Video Team (JVT) jointly established by ITU-T and ISO / IEC. H.264 still uses a hybrid coding scheme combining prediction and transformation. In order to obtain higher video compression coding performance and wider applicability under the same coding framework, the H.264 standard introduces many new technologies, such as 1/4, Motion estimation with 1/8 pixel accuracy, inter prediction with multiple reference frames, intra prediction, loop filtering, adaptive arithmetic coding, etc. The H.264 video coding standard has significantly improved coding quality and compression ratio compared with the original video coding standard.
On the same visual perception quality, the coding efficiency is 50% higher than the previous coding method. The encoding performance of the H.264 standard surpasses all previous video encoding standards and has good application prospects. It is widely used in video compression and video surveillance.
At present, the research of the H.264 codec standard is mainly divided into two categories: algorithm research and hardware implementation. The hardware implementation schemes are mainly divided into three types:
(1) Scheme based on PC platform. This program implements coding and decoding for pure software, and uses MMX and SSE / SSE2 and other multimedia instruction sets to optimize the program. It has the advantages of low development cost and short cycle. The CPU architecture of the PC is not suitable for processing digital signals, so the effective utilization rate of the CPU is relatively low.
(2) Pure hardware solution based on ASIC chip. This solution solidifies the video codec algorithm into hardware, which has the advantages of high integration and short development cycle. However, the special type is relatively strong and the product is not easy to upgrade. H.264 codec chips have appeared on the market, such as Fujitsu's MB86H51, Hisilcon's GOALTM Hi3510, and JVC's JCY0237 LSI.
(3) DSP-based combination of software and hardware. This scheme uses DSP chips and other peripheral chips to form a processing system, which has the advantages of high development flexibility, strong processing capacity, low development cycle, low power consumption and easy upgrade. With the continuous improvement of the cost performance of DSP, this solution has become the ideal solution for the current H.264 encoder hardware implementation.
The H.264 codec standard has the characteristics of high compression ratio, wide adaptability, strong fault tolerance and high image recovery quality, etc. It has good application prospects in real-time systems. TMS320DM642 is a digital multimedia processing chip launched by TI for video and image processing applications. It has the characteristics of strong processing capability and high integration. It is one of the ideal chips for implementing H.264 video encoders. Many domestic and foreign companies are developing or have developed a video surveillance system based on DM642.
1 X.264 encoder transplantation
X.264 was initiated by the Central Institute of the Paris Central School in France in June 2004. It is a project completed by many video enthusiasts. It focuses on practicality and strives to reduce encoding calculations without significantly reducing encoding performance. The complexity eliminates some modules in JM that are relatively time-consuming but have little impact on performance, such as multiple reference frames, unnecessary block modes in inter-frame prediction, and CABAC. The X.264 encoder utilizes multimedia hardware acceleration instructions based on the X86 architecture such as MMX / SSE / SSE2 in the program structure. It is necessary to mask the relevant X86 instructions and simplify some functions to make it simple in structure and easy to execute on the DSP. After the simple transplantation of the X.264 encoder, the encoding efficiency on the DM642 platform is extremely low. Table 1 shows the encoding results of the transplanted X.264 encoder on the DM642 platform.
It can be seen from the results that the coding efficiency of X.264 on the DM642 platform after transplantation is very low, which can only reach an average coding rate of 0.6 f / s. It is necessary to further optimize the characteristics of the X.264 encoder and DM642 to Improve coding efficiency.
Shenzhen ChengRong Technology Co.,Ltd. , https://www.dglaptopstandsupplier.com