Aware of video surveillance cameras allow the future can be big data
Editor:admin Date:2015/8/8 Click
Video surveillance in high-definition, after intelligence, resulting in vast amounts of video and image data, and then the traditional back-end storage and intelligent analysis brings a series of problems. In the video surveillance industry applications, application start landing big data implementation, but the front-to-back big data applications, is a complete closed-loop big data technology applications, can now propose a total solution security vendors, very few.
Of
Shortcomings of the traditional big data technology applications
To find shocking "8.10 Chongqing shooting robbery," the suspect Zhou Kehua in the video monitor screen, the local public security department spent about two thousand police in the back-end video surveillance for up to ten hours a day Figure Search, total video views equivalent to 830,000 films, spent a lot of manpower and resources. With Big Data technology popularization and application in the field of security, based on the back-end intelligent analysis of large data server technology began to be applied to the public security sector, to a certain extent, to achieve the goal of finding a computer-based functions.
From the technical analysis, this solution enables intelligent analysis, but commercial application point of view to see, at the time of the massive high-definition video intelligent analysis of the back-end server hardware configuration, process performance requirements are very high, user cost will be greatly increased. A blade server can only analyze a few high definition video, and will cost tens of thousands, leading the way to do video analysis, it is necessary to increase the cost of nearly a million. Such a high cost for public security, transportation and other industries in terms of thousands of cameras installed to achieve large-scale application of intelligent analysis of the pressure is very great.
Big data front to back closed-loop applications
In this regard, Suzhou Keda Technology Co., Ltd. believes that before the advent of perception camera, for video surveillance applications, large data collection techniques from the front to the middle of the store, to the back-end application, are not well closed loop. In fact, before the advent of perception cameras, video surveillance big data applications only stay in the theoretical stage.
To solve the problem of video data mass storage and applications, make big data technologies to better serve the public security sector, the best solution is to move to the back-end intelligence analysis function front of the camera before. Namely the use of an image recognition camera, awareness and generate three types of data collection: unstructured video data, image data, and structured text data, and traditional video surveillance cameras only generate video data.
This is a surveillance video at the depth of large data applications, the camera technology proposed new intelligent analysis concept: the intelligent analysis capabilities into the front of the camera inside, let the camera do the identification of producing an image data. As a simple example, if the sensing cameras installed in the square, so any passing people and cars will be snapped off, and then the basic characteristics of the human face features, clothing color, model, body color, license plate and describe storing text. That perception cameras in addition to capture and output video, pictures and features will be generated video inside a moving object description text of these two data. These vast amounts of HD video data and pictures stored in the cloud storage, the text information is stored in a large database back-end servers, data storage between the two is associated with an index. For example, when searching for "red Mazda," the keyword, the system will also provide all the red Mazda vehicle through which the designated place, and also the associated corresponding video.
Video surveillance of large data possible future
Aware of the camera combined with the application of large data, video surveillance and knocked on the door between big data, so that businesses can achieve by its image-based big data retrieval, analysis and depth of application. In specific applications, as of the end of 2014, Kodak has built a perception based front-end, back-end big data applications Guangxi, Fujian, Liaoning, Gansu and other places safe city. A Case Study, Nanning "Skynet" system is the first case of an application-aware Kodak camera. At present, it has been on the line for about 3,000 Kodak cameras perception, accounting for Nanning "Skynet" front-end the total number of nearly 1/3. Among them, the basic features of the identification of people, vehicles Category / color / direction and other main characteristics of more than 2,000 cameras to identify vehicles, car standard, body color-based vehicle details vehicle mount camcorder about 600 units, covering important road junctions Nanning, squares, key units entrances and other major security monitoring sites. The 3,000 road-aware applications means that each individual cameras, each vehicle photos and detailed text information Full Nanning all important security sites can be stored for up to several years, which means more than 300 in the past to the aid of intelligent server to the high cost of doing video analysis equipment can be omitted, but also means that urban safety management and truly entered the era of big data have practical significance.
In the IT era steering DT era background, the availability of secondary processing and application of big data and its value is being recognized by the people. In the field of security, security for large data applications are integrated force change, the video data as unstructured data in computer information applications has been a hot potato. Kodak aware of the camera appears, make security video surveillance of large data into the actual operating level, that perception camera in front-end collection, analysis, identification, provide valid data to back-end, big data platform to cloud the way the video data storage, secondary depth analysis, forecasting judgment result, so as public security, transportation and other industries from the front to the rear of the closed-loop applications platform in massive high-definition video data dawn.
