Microwave Video Monitoring System Based on Da Vinci Platform

Microwave Video Monitoring System Based on Da Vinci Platform

[Date: 2008-7-18]

Source: China Power Grid Author: Zhao Yong

[Font: large, medium, and small]

　

 

0

The electric power system is the main artery of the national economy and the most basic and essential guarantee for the normal life of the people. Moreover, with the development of modern society, people are increasingly dependent on electricity. Therefore, it is necessary to ensure the normal operation of the power system.

To ensure the safe operation of power facilities, a video image monitoring system is required. Because microwave transmission channels have been installed on the microwave station, the existing microwave channels can be used for local materials.

Microwave channels have narrow bandwidth and high error rate. H.264 adopts the world's most advanced compression standard to form a high-quality image monitoring system. To use this latest video compression technology, the Da Vinci platform was launched by Texas Instrument (TI. It contains two cores, one ti tms320c64x + and one arm9-core. One is responsible for operating system control and scheduling tasks, and the other is responsible for processing computing intensive tasks. The two cores work together to form a powerful multimedia processing platform.

Based on the dvevm hardware of Texas instrument, we have determined the hardware structure of the video transmission system, designed and developed the software for video terminals and the software for receiving and browsing videos on the central station, completed the principle development of the entire system. The final performance test shows that we can transmit high-quality CIF images with a bandwidth of less than 64 KB. With a bandwidth of K, the video images with high resolution D1 can be continuously transmitted.

1. Microwave-Based Video Monitoring System Structure

The basic mechanism of the Image Monitoring System for microwave stations is as follows. It consists of several video transmission terminals and a central station. The video information of each video monitoring site is transmitted to the computer on the central site through a point-to-point Microwave Channel for monitoring. The key device for microwave transmission is the Alcatel a9800, which is a point-to-multiple-point digital wireless access system that uses the spatial interface of TDM/tdma fdd for transmission. The system principle structure 1 is shown in.

2. hardware platform of the video terminal da Vinci davincitm

The structure of the Da Vinci-Based Video Terminal 2 is shown in. It mainly refers to the hardware system based on Da Vinci dvevm (Digital Image Evaluation System) introduced by TI:

DVSDK: a digital video development platform provided by Monta Vista. It includes:

(1) eXpress Configure Kit: it can turn different software modules into an executable file, avoiding manual integration of software including ARM and DSP, and how to coordinate their work. For example, Audio/Video Codec Module, TI Codec Engine, ti dsp/BIOS real-time kernel, and ti dsp/BIOS Linker.

(2) TMS320C644x SoC Analyzer. It is a single graphical system that helps developers identify system operation bottlenecks, identify problems, and solve them. It includes system integration, load distribution, data input and output, and other behaviors.

(3) Monta Vista operating system: Monta Vista is recognized as a very stable Linux operating system. However, DVSDK has made a lot of optimizations for digital video applications, make it the best operating system to support video processing.

3 Application of video compression technology H.264

The basic process of H.264 is that the encoder first splits the image into an image and then divides the image into a macro block. Each macro block is processed based on the frame type. For independent (I) frames, the so-called intra-frame prediction is used. For non-independent frames, the inter-frame prediction, that is, the so-called motion search, is used for prediction. DCT transformation is used for prediction, and entropy encoding (arithmetic or variable code length encoding) is used at last ). H.264 the compression ratio has been greatly improved due to the use of the following technologies:

(1) 1/4 and 1/8 motion search technologies improve the matching accuracy of motion search;
(2) Multi-reference frame technology;
(3) fine-grained prediction technology within a frame;
(4) 4 × 4 small pieces of prediction technology, making it easier to match the block.

On the basis of H.264, we have also made the following improvements:

(1) apply the visual model to further reduce the bandwidth occupied by video signals. That is to say, a larger quantitative step is used for the larger part of the transformed image, while a smaller quantitative step is used for the smoother part. There is almost no obvious difference between the compressed image quality and the image without a visual model, or only a slight decrease, but the image compression ratio can increase by about 10%. However, the difference in the SNR is slightly greater, but for many applications, people are not very concerned about the specific SNR size. What we see as reality prevails.

Assume that the quantization level of the benchmark is Q. The variation of the block to be transformed ranges from the absolute error of the adjacent horizontal pixels to △:

Formula: n is the fast size (4 or 8), P (I, j) is the pixel value at this position. The New Quantization level of the block will be adjusted:

Formula: qmin, Qmax is the set minimum and maximum quantization level (10, 51), floor indicates the Integer Operation, μ is the adjustment coefficient. After compressing the mpeg Test Series News (CIF), the following results are displayed: although the SNR is slightly different, the visual difference is not very large, as shown in comparison 3.

(1) Double-finger motion search. Almost all current motion searches use a single indicator, that is, absolute error and sad. In many DSPs, There are instructions for finding sad, such as the C6000 series of Ti. In fact, the application of double indicators is more effective in motion search. In addition, the location of the matching module can be found more accurately to improve the encoding efficiency. More importantly, double-indicator motion search makes it easier to intelligently determine whether a small block is moved from a small block in the previous image. Therefore, coding is no longer required to improve the coding efficiency. The effect is shown in Figure 4.

(3) Multi-Resolution Double-indicator motion search using wavelet transform (Lifing-scheme, first, obtain an approximate description of the motion vector at the largest resolution, and then gradually refine the motion vector in the high-resolution image so that the motion vector can be quickly found, avoid local minimum values, as shown in Software Interface 5.

4 Conclusion

Based on the DVEVM hardware of Texas instrument, we designed the video transmission system. Based on the DVEVM hardware of Texas instrument, we designed the hardware structure of the video transmission system, the software for video terminals and the software for receiving videos from the central station have been developed to complete the development of the entire system. Performance tests show that CIF image monitoring can be implemented in a 64 K bandwidth environment, and the monitoring frame rate can reach 2 ~ About 5 frames, with a latency of 2 ~ 3 s to meet the basic requirements for monitoring purposes. If you bind a 6*64 K = 384 K channel, you can use a bandwidth of less than 384 K, complete the continuous transmission of high-quality D1 (704x576) images with high resolution and achieve the design requirements. The average SNR of image quality is about 30 dB, which can meet the visual needs and identify criminals.