IP video surveillance system (VSIP), hardware processing of network traffic is part of the camera. This is because the camera video digital signal, and in order to overcome the network bandwidth constraints, signal transmission to the video server before compression. Different type, such as DSP microprocessor architecture, is GPP/maximum performance of hope. Interrupt intensive tasks, such as video capture and storage and can be assigned to GPP video, and video compression by DSP the MIPS dense. Data transmission to the video server, server to file storage to the hard disk video compression, which can overcome the storage device simulation of traditional performance degradation.

For digital video signal compression has developed a variety of standard and can be classified into the following two categories:

Static image compression - independent video coding each frame as a static images. The standard is the most famous JPEG. MJPEG JPEG arithmetic coding standard with each frame.

Figure 1 shows the h.264 encoder diagram. Similar to the other yankees video coding standard, ME it on macro unit (16 (16 pixels) with each frame macroblockMB unit (macro). It has a prior to access and reconstruction. Prior to access code for a frame. Reconstruction from coding bits produces a reference frame.

In prior to access (DCT), to Q within each MB can model or interactive mode. In the model, by ME, for reference in earlier module code. MB In the interactive mode, formed from now sampling frame.

IDCT to rebuild pathway (IQ) are designed to ensure encoder and decoder for the same reference frame to produce an image. In addition, encoder and decoder of error between the accumulated.

Due to the video sequence contains many related information often frame, so the yankees method can achieve higher mind compression ratio. For example, 30f resolution (standard for NTSC/s) h.264 encoder, video coding, 2mbps can reach 60:1 ratio of the image compression ratio. In order to achieve similar quality, MJPEG compression ratio approximately 7 ~ for now.

Compared with ME MJPEG yankees have several advantages. The main advantage, JPEG need very little calculation and power consumption. In addition, most of the PC has software can decode and display JPEG images. In a single image or a special events recorded images (such as someone into the MJPEG) is more effective. If you cannot guarantee network bandwidth, it is MJPEG selection. By ME, 1 frame, the delay/GOP will result in the loss of delay/lost, this is a decoding frame will not be available until the first reference frame.

In a typical digital surveillance systems, video capture from video compression, sensors, and data flow to video server. Don't interrupt on DSP structure to achieve the video encoder task, it is because each between switch contains a number of registers storage and ultra-high speed buffer switch. Therefore, the different type of structure is ideal, video capture and data flow tasks can take off in digital signal processor (DSP). In video surveillance in below shows the application example. GPP/DSP processors

DSP/GPP SoC of video signal digital compression, developers first should be properly distribution function module to achieve good performance of the system.

EMAC drives, TCP/IP stack and HTTP server (and flow to the external compression video signal, and working together in the ARM ATA drive, this will help realize off. In DSP, Compression in DSP core should be realized in nuclear, this is because the VLIW framework to handle such computation intensive tasks is very good.

Once the video input through the processor from cameras captured video frame mouth video encoder, the implementation of the original video compression, then compressed video storage to the hard disk.

In the application of this system, and can be used in real-time video surveillance PC, this is on the video retrieval of data flow and server decoding, show on the monitor data stream. In this case, through the Internet can board retrieval 1 coding JPEG image files. In the single PC can monitor multiple data stream. From the multi-point network can also monitor data flow at. Different from the traditional simulation system, VSIP central office can via TCP/IP network video server and can even physical location of any place. This failure of single point into digital camera, instead of the central office. Also can dynamically configures JPEG image quality.

Related Product: