Intelligent video (IV, Intelligent Video) is derived from computer vision (CV, Computer Vision) technology (computer vision technology is one of the branches of artificial intelligence research), it is to establish a relationship between images and image descriptions, so that computers can Digital image processing and analysis to understand the content in the video screen, to achieve the purpose of automatically analyzing and extracting key information from the video source, that is, intelligent video analysis technology (IVS).
Fault diagnosis in the monitoring system
Since the birth of intelligent video analysis technology in the 1990s, after decades of development, this technology, which originated from computer vision, is gradually gaining widespread attention along with the gradual application of commercialization. Some professional video analysis and research manufacturers at home and abroad have successively launched products of various forms, such as intelligent video servers, intelligent network cameras, intelligent analysis hard disk recorders, and intelligent video analysis software. As a high-end application of video surveillance, functions such as perimeter detection, behavior analysis, and video fault diagnosis have been successfully applied in key industries and have gradually shown their power. Take the safe city monitoring system, on the one hand, it is mainly reflected in some important roads, communities, public places, etc., to monitor and alarm suspicious targets that appear through video surveillance. On the other hand, it focuses on the later operation management process of the monitoring system to detect the common faults of the front-end camera and the low quality of the video image through video analysis technology to achieve effective maintenance of the monitoring system.
As an innovative product in the security field, the video quality diagnostic system is a typical application of video analysis technology in the operation and maintenance of safe city monitoring systems, and is also a relatively common product. It is mainly used in the control center of the large-scale monitoring system. By controlling the video switching output of the matrix host of the monitoring center or connecting the digital video streaming media management server to obtain the video signals of all the front-end cameras, the snowflakes, scrolling, blurring and partial appearance of the video image Color, picture freeze, gain imbalance and pan / tilt out of control common camera failures, as well as malicious obstruction and destruction of surveillance equipment to make accurate judgments and send alarm messages; video surveillance equipment is increasing today, its application in surveillance systems It will definitely help users to quickly control the operation of front-end equipment and easily maintain large-scale security systems.
Core technology of video quality diagnosis
The video quality diagnosis system adopts the method of video image analysis to detect various common faults of video in the monitoring system. Judging from the types of camera failures that are now common, there are many factors that affect the video quality of video surveillance systems. The main points are summarized as follows:
· Incorrect camera settings or aging and failure of devices, including camera resolution, camera sensitivity to light, lens focus adjustment, color correction, etc .;
· Video signals in large-scale surveillance networks are transmitted through long-distance cables, multi-level matrix switching and multi-level network forwarding. Power supply, controller and other interference signals may cause strong interference to the video signal, line aging, loose connectors, etc. Changes may cause video noise;
· A large number of PTZ dome cameras are used, and long-term sports zoom may cause some dome cameras to have wrong orientation and uncontrollable faults.
In view of the various video failures mentioned above, the failure types can be divided into 8 types: lack of video signal, abnormal video definition, abnormal video brightness, video noise, video snowflakes, video color cast, picture freeze, and PTZ motion runaway. Among them, two failures, video signal loss and picture freezing, can be concluded by manually designing a method based on video image comparison; PTZ motion out of control is a motion instruction issued by the fault detection system, and then detected by motion analysis of the video image Whether there is a fault; and for the other five types of faults, it is difficult to detect by manually setting rules. This requires machine learning to allow the machine to simulate human visual response and detect whether the video is faulty.
In response to these five different types of video failures, five different machine learning-based detectors are designed. Each detector is responsible for analyzing whether a certain type of video has a certain failure and the severity of such failure.
In the actual video surveillance system, a large number of video clips are extracted, including normal videos and videos with various faults, training samples are formed, and human visual characteristics are simulated, and a large number of video image feature parameters are extracted for different fault types to Trained to get detectors that diagnose different faults. In the analysis stage, obtain a fixed-length video that needs to be analyzed, use different fault detectors to extract the corresponding video image features according to the detection items set by the user, and then input it into the trained fault detection model , You can get the fault evaluation result of the video.
Based on excellent underlying algorithms, the video quality diagnosis system has the following technical characteristics:
High accuracy: A large number of actual video surveillance system videos are used as training samples. Various fault detectors are derived from the actual system and have been tested by a large number of actual systems, so the detection accuracy is high;
· Good camera angle adaptability: The training samples of the fault detector come from many different scenes, covering many common camera monitoring angles in the security video surveillance system, so they are good for various camera angles, focal lengths and different camera content Adaptability;
· Unique ability to resist the movement of the ball machine: In the design and training process of each type of fault detector, the changes of video image characteristics that may be brought by the movement of the camera head and the zooming of the lens are considered, During the detection process, the camera motion analysis is first performed. Once the camera is found to be in the PTZ motion process, it is no longer detected whether the PTZ motion is abnormal to prevent the motion command from being sent to affect the current dome motion; Perform motion-insensitive features for other types of fault analysis to avoid false alarms or false alarms due to motion;
· Excellent environment adaptability: The algorithm module is not sensitive to the changes of light and shadow caused by traffic flow, pedestrian flow, season and climate in the scene, so it can be applied to many different outdoor environments;
Reinforcement learning ability: There is still a clear gap between the existing video quality diagnosis system and human fault recognition ability, so the difference in application scenarios has an impact on the performance of the video quality diagnosis system. Like the human visual system, the video quality diagnostic analysis module also has acquired reinforcement learning capabilities. As long as new local samples are added to retrain the detector, the performance of the algorithm will be further improved.
Double Burner with Built in Hob
Gas Stove 2 Burner,Double Burner Gas Cookers,Brass Burner Stoves,Double Burner Gas Stove
Xunda Science & Technology Group Co.ltd , https://www.xundatec.com