Nowadays, video surveillance systems have developed various benefits in terms of compression and video quality, to the point of allowing us to see the smallest details in the image.
By Juan Carlos Yáñez*
However, this is not enough to optimize the surveillance task itself. Several studies reveal that the task of a video security operator is effective for a limited period, verifying that the attention in this type of work greater than 4 hours becomes difficult and can produce errors of appreciation. These studies show that it is not enough to have a high-quality image and resolution, but that intelligent video analytics are also necessary that allow operators to benefit from the necessary tools to perform a better security work.
The development of various intelligent video analytics or allows security camera operators to optimize their response times to an alert and / or event captured by the camera. An intelligent video analytics event can generate various responses in the VMS system such as pop-ups with the live video image or playback of the event on the videowall, sending an email, issuing an audible alert on the camera itself, among others. This results in the reduced need to view all cameras at the same time and only focus on the most important video images and respond to alerts generated by intelligent video analytics.
In addition to detecting various behaviors, intelligent video analytics must minimize false alarms through intelligent algorithms that allow discriminating constant or recurrent movements or behaviors such as the movement of tree leaves or lighting changes during the day or change from day to night.
The video surveillance industry no longer only focuses on video security itself, but also on the observation of the behavior of people in a scenario such as an industrial plant, retail stores, shopping centers or customer service offices. To this end, intelligent video analytics have been developed such as:
- People counting
- Hot zone map (heat maps)
- Row management (lines/queues)
The people count feature tracks the number of people entering and leaving a specific area, and provides statistics along a timeline. On the other hand, the heat map function reflects the movement rate of people in the scene, thus helping to recognize patterns of people's movement in the scene.
Like people counting and heat maps, queue management allows you to calculate approximately how many people stay in a specific area for a certain time, so you can check real-time data as well as daily or weekly statistics.
In addition to the intelligent analytics of video for security such as: motion detection, face detection, abandoned / removed object, etc. The industry has developed others such as:
People prowling
Loitering detection is based on the observation of the permanence for a certain time of an object in a defined area. This intelligent video analytics is based on the intensity of movements of an object, intensity that must be included within a range configured by the user.
Mist detection
The mist is composed of a set of particles of liquid or solid water, falling or suspended in the atmosphere. Because these particles are larger than the particles in the air, they cause light to scatter more. The haze detection algorithm is based on the detection of a blurry image in the scene, which in turn automatically detects the haze and sends an alert to the video management system. In turn, the cameras have the option of automatic anti-fog, manual or in response to the event of fog detection, allowing to adjust the contrast in certain parts of the scene, based on combined alpha images with greater contrast applied to the entire image.
One of the most recently developed intelligent analytics is audio classification, which is based on audio detection, extracts the characteristics of the audio source captured through the internal or external microphone and calculates its proximity based on the data stored in the system, after which it selects the sound closest to the one captured and generates the event. The algorithm used can classify sounds into screams, gunshots, explosions and glass breaks.
In addition to the intelligent video analytics mentioned, video surveillance cameras have other analytics such as automatic digital tracking, blur detection, intrusion detection, among others.
With the use of intelligent audio and video analytics technology, a single operator will be able to monitor several video devices, thus reducing the percentage of critical events that could happen.
*Juan Carlos Yáñez is a sales application engineer for Peru and Bolivia at Hanwha Techwin.
