Tag: Video quality Hardware compatible
This article refers to the address: http://
Hardware processing and compatibility issues
Under the demand of various projects, video quality diagnosis technology is increasingly concerned by manufacturers and users. Nowadays, there are not many manufacturers focusing on video diagnosis and development. Nearly manufacturers of intelligent analysis/platform development support video diagnostic functions more or less. While customers have a desire to have more features in their surveillance systems, more features mean more investment. These reasons lead to a lot of problems in the growing video diagnostic technology to be improved.
Limited hardware processing performance
No matter how good the software algorithm is, it is based on hardware implementation, hardware processing capability can not keep up, and then powerful software processing functions can not be fully implemented. For now, the primary issue that constrains video quality diagnostic analysis is hardware. Dongfang Netpower IVS Product Department colleagues pointed out that video diagnosis is a complicated process. First, the diagnosis process requires a large amount of system resources. This implementation mode will put a lot of pressure on the back-end resources, especially with high-definition images. The increase in the processing capacity of the server is more dependent; Second, subject to network bandwidth constraints, the IPC system diagnostic tool is based on PC-based back-end software analysis, because the network needs to transmit video to the back-end, may bring lost Package and other issues. At present, Dongfang Netpower has developed a detector based on 4-core concurrent 4-way diagnostics, which can diagnose 3,600 channels per hour. In terms of detection speed, when there is no PTZ detection, the detection time is less than 2.5 seconds, and when there is PTZ detection, the detection time is less than 4 seconds. To solve the simultaneous operation of various algorithms, they adopt the load balancing principle, that is, integrate multiple detections in one system. (Detection module), the system will distribute the video stream to the idle detector according to the real-time load capacity of the system for diagnosis, and achieve the purpose of rapid detection and full utilization of system resources.
At the same time, the industry is optimistic about the processing performance of hardware, and believes that the existence of the problem is temporary. With the replacement of the processor, the limitation of the computing power of the server will be solved; the problem of network transmission will also follow the communication. The development of technology has gradually eased.
Compatible issues
Projects that have an urgent need for video diagnosis are basically large projects, and equipment provided by multiple manufacturers is a common phenomenon for large projects. At present, the compatibility between different devices has not been resolved, and this situation will have a major impact on video diagnosis. The vice president of Hangzhou Zhinuo Ying pointed out: "The current technical difficulties mainly lie in the coding equipment of different manufacturers at the front end. Different types of IPC devices are diverse and there are compatibility issues."
The status quo of the industry's respective policies has a profound impact on the smooth development and application of the project, so users, system integrators, and engineering companies should be cautious when selecting diagnostic products. As an R&D manufacturer and system integrator, Beijing Yingsijie, his engineer Wang Huiwu told the author: "In a large-scale monitoring system, there may be many manufacturers' products. At this time, we will consider the compatibility of many products. This is an important factor in the selection of diagnostic products, including hardware devices and diagnostic algorithms."
Judging criteria are not uniform
There is still some controversy between the industry on the judgment of whether the video has quality problems. The first is the setting of the software identification standard. Different manufacturers have their own standards when developing diagnostic algorithms. The judgment criteria for video quality are inconsistent, and the diagnosis results for the same video may also differ. In order to solve this problem, some video diagnostic R&D manufacturers have made flexible developments in diagnostic systems, such as Wen'an, Dongfang Netpower, and Zhinuo Inte. Let's listen to the introduction of Beijing Yingsijie Wang Huiwu: "There is no obvious difference in the effect of the video quality diagnosis system, so we do not have a fixed video quality diagnostic system supplier. In specific applications, (different products) The same detection threshold to detect may have different results. Therefore, the video quality diagnosis system can only be applied as a reference system in the existing system construction.†Secondly, the definition of human judgment, different people, they are The definition of video quality issues made by the slight picture quality of the same video will be different.
Monitoring perfection and night vision effects
Very rare, the detection of video quality is not perfect.
The existing diagnostic systems are mostly for the detection of common video quality problems. Through long-term development and application, the recognition rate is well guaranteed. Du Jianying, general manager of Abbott Technology (Beijing) Co., Ltd. believes: "For some basic functions. It should be said that the accuracy can reach more than 90%." But for some rare video quality problems, such as a thin white line in the middle, video ghosting, shadows, partial black screen, partial stripes, menu subtitles, etc., the system for such problems Either the detection module does not have such detection function, or the detection accuracy is very low, or even not detected. Hikvision (002415, shares it) Huang Danping pointed out: "These problems only R & D manufacturers continue to collect a variety of different problem models, and constantly expand the type of detection to gradually improve the detection function of the video diagnostic system."
Algorithm expectation optimization
As the core of the system, the algorithm is the key to the success of video diagnostic applications, but there are still two major difficulties:
Recognition accuracy: Diagnostic system relies on a rich database for image quality diagnosis. Abbott Du Jianying believes: "The reason why the recognition accuracy needs to be improved is because the image processing algorithm is relatively simple, and the image is learned by performing the "image understanding" technique. The meaning of moving and stationary targets, describing the normal state of various targets in the picture, and then comparing them through the texture, edge, etc. of the target, so as to better realize the analysis.†Thus, the capacity of the database and the optimization of the comparison algorithm are known. The degree is the basis for determining whether the system is judged correctly or not. In addition, subject to the current hardware problems, there are not many algorithms for concurrent processing by a single identification module. For example, there are no more than 10 algorithms for concurrent recognition of the identifiers of Zhinuotech. Only by optimizing the algorithm can the hardware of the recognition system be reduced. Over-reliance on capabilities, so that a single device can run more algorithms concurrently and improve recognition accuracy;
Recognition rate: According to Ying Sijie Wang Huiwu, the diagnostic system used by them is about 2-4 seconds for a single device. It takes more than one hour to complete the round-trip diagnosis of a 1000 camera. The more cameras, the more cameras The longer it takes, it is not conducive to the application of large projects. The currently common solution is to use multiple detection servers simultaneously.
Poor diagnosis at night
Due to the black-and-white picture at night and the influence of nighttime lighting, it is difficult for the video diagnostic system to make accurate judgments on the picture problem, even if it is caused by a camera or an environmental cause. Hikvision Huang Danping believes that the main solution to this problem is to detect the time setting to bypass the night, because the detection frequency is generally once a day, basically the detection time is set in the daytime, and the camera video quality problem can be detected more accurately.
Steel Light Pole, Steel Street Light Poles, Stainless Steel Light Poles
YIXING FUTAO METAL STRUCTURAL UNIT CO.,LTD( YIXING HONGSHENGYUAN ELECTRIC POWER FACILITIES CO.,LTD.) , http://www.chinasteelpole.com