Intelligent video analysis of events in IP video surveillance systems.
Currently, much is said about the advantages of integrated security systems. However, traditional approaches to integration often imply attempts by integrators to combine systems from different manufacturers that successfully perform their functions separately. In this case, as a rule, many problems arise with product compatibility, software version support, etc.
The modern trend in providing comprehensive security for a facility is when the customer chooses solutions from one manufacturer for the purpose of coordinated operation of systems and their unification into an integrated security system with centralized management. Thus, the customer has relations with one manufacturer across the entire spectrum of equipment: from ACS, video surveillance systems to perimeter and territory protection radars for facilities.
Domestic and foreign manufacturers of security systems offer universal IP video surveillance systems, which are built on a modular principle and can be integrated with radar security systems, access control systems, fire alarms, license plate recognition systems, any sensors, etc. Such systems implement a full range of video surveillance capabilities: from a simple video recorder — a system in which the user gets the opportunity to view «live» video, record and play back the archive, to a video surveillance system with computer vision with the ability to recognize potentially dangerous situations, search the archive by events, tune out false alarms, etc.The universal functionality of IP video surveillance systems has freed users from the work of combining solutions from different manufacturers. Thanks to the advent of universal IP systems, users have the opportunity to equip an object with any type of video surveillance using a single universal platform. Now the user independently forms the functionality of the system and the required number of video channels to solve specific problems. This eliminates overpayment for unclaimed functions and «extra» video channels.
Universal IP video surveillance systems support several types of video channels, which can be divided as follows:
channel with video recording functions;
channel with video analytics functions (motion detection using IP video cameras);
channel with advanced video analytics functions (video surveillance systems with computer vision).
By choosing channels with video recording functions, the user receives a fully functional video surveillance system with the ability to view live video, record and play back archives, and connect PTZ cameras and control them manually.
Video channels with video analytics functions (motion detection via IP video cameras) allow you to create a video surveillance system with the ability to detect moving objects, display their movement trajectory and transmit information about penetration into the protected area in real time to the operator in automatic mode.
Users have the option of connecting PTZ video cameras with manual control. The PTZ camera can be aimed based on the target designation of a stationary camera.
Video channels with advanced video analytics allow you to create a video surveillance system with computer vision. Video surveillance systems with computer vision are designed for round-the-clock, all-weather protection of the perimeter, the territory of objects and approaches to them by automatically detecting and recognizing targets (people and cars) and potentially dangerous situations (an abandoned object, fire, smoke, an attempt to block the camera, the appearance of objects in the alarm zone, a thrown object, etc.) with the transmission of an alarm signal and video information to the operator in real time. Due to this, even the largest protected area or object equipped with hundreds of video cameras does not require a large staff of security personnel. Video surveillance systems with computer vision operate on the principle: «one object — one operator».
Video surveillance systems with computer vision are in demand for ensuring fire safety. Thanks to built-in video fire and smoke detectors, video surveillance systems recognize the occurrence of fires and smoke in real time. Now the operator receives not only an alarm signal for fire-related events, but also automatically receives a video image of this situation, which will increase the reliability of detecting fires and preventing them at an early stage.
Detecting fire and smoke using video analytics algorithms has a number of advantages over traditional methods.
The most common methods of detecting smoke and fire are fire detectors that react to smoke, flame, etc. Combined fire detectors that detect fire based on several factors are widely used.
The disadvantages of detectors are that fire detection is usually carried out indoors, and the detected factors (smoke, fire, gas, etc.) must naturally reach the sensors. This significantly increases the time it takes to detect a fire. In addition, it is extremely difficult for security personnel or the Ministry of Emergency Situations to make decisions based only on signals from detectors, which can generate false alarms or not work at all. There are also numerous cases where sensors issue an alarm signal with a long delay.
The use of fire detectors is not reliable for objects located outdoors or at objects using natural or artificial ventilation, such as in tunnels.
Video surveillance systems with video analytics allow detecting smoke and fire even in open spaces, where traditional sensor systems are not always effective. Fire and smoke are detected at the moment of ignition, and video information allows for accurate and real-time determination of the location of the fire and its size.
IP video surveillance systems with computer vision can be integrated with any fire systems and detectors (heat, smoke, gas, flame, etc.). In this case, the operator is automatically sent alarm signals and video information from the site of a fire or smoke when a fire detector and/or video surveillance system is triggered.
The integrated system can also issue an alarm signal and video information when, for example, only a fire detector is triggered or only when fire and smoke are detected by the video surveillance system, as well as when both systems are triggered together.
Thus, video analytics can be used in all areas of life where there is a risk of fire.
Integrated security systems based on universal IP video servers allow for comprehensive protection of perimeters and territories of facilities with the possibility of centralized multi-user control of security system functions, including in autonomous mode.
The implementation of integrated systems allows combining security systems of different levels (video surveillance, access control, sensors, fire protection, etc.) into a single security complex with centralized control of the situation and the ability to promptly interact with law enforcement agencies, security centers, the Ministry of Emergency Situations and other emergency services.
IP video surveillance systems built on a modular principle are capable of ensuring the security of both small facilities and large, geographically widespread enterprises.
Intelligent IP video surveillance systems are characterized by the following functions:
real-time recognition of targets (people, cars) and potentially dangerous situations, such as fire, smoke, abandoned object, a person stopping in an alarm zone, a thrown object, vandalism towards the camera (painting over the glass of the casing or lens, blocking the camera lens);
all events (sensor triggering, target appearance, fire, smoke, abandoned object, etc.) are displayed on the object map;
automatic targeting and tracking by rotating video cameras;
rejection of false alarms due to illumination by car headlights, insects appearing in the camera's field of view, tree swaying, camera shake, precipitation, cloud shadows, etc.;
control of the system operator's actions, which eliminates the influence of the human factor, collusion between operators and violators;
self-diagnostics of operability with the issuance of an alarm message in the event of failure of system elements;
the ability to search the archive by time, target class, camera number, event, operator comment;
interactive display of video information via a map (clicking on the map displays a video image on the screen).
The main problem of existing video surveillance systems (analog, digital, IP) is the need to work with almost every camera of a separate operator. Otherwise, the video surveillance system becomes either meaningless or is limited to archiving huge video streams without the ability to process them in real time. This statement is true even if the cameras are installed in deserted rooms: a person is not able to hold attention for long in the absence of events, and saturated scenes quickly put him into a stupor. A person is a bad observer, he cannot cope with more than one camera.
Video surveillance systems with video analytics automatically, without human intervention, analyze the video stream and find in it important information for the consumer, which is many times less than in the entire video stream.
Thus, video surveillance systems with video analytics allow you to solve the main problem of video surveillance: now one operator can service an order of magnitude more cameras than with conventional video surveillance. This is achieved due to the fact that the operator receives information about potentially dangerous situations in real time automatically after the automatic analysis of the video stream by the video surveillance system.