Cloud computing and cloud storage boost video surveillance development

The diversity of market demand provides market space for various storage solutions. Therefore, in the future for a considerable period of time, DVR, NVR, sdi-dvr and Fcsan, Ipsan and other storage solutions will coexist. Cloud computing and cloud storage are likely to have a huge impact on the video surveillance market over the next few years and push the video surveillance industry to a new, higher stage of development.

Improve Storage network performance

Distributed File System SDFs (Skydistributedfilesystem) meets the new requirements of security video surveillance, solves the problem of storage and retrieval of massive video data in the city, and is scalable for large-scale data-intensive applications with video, picture file storage and management as the core.

Redundancy in most block replicas

Highly reliable, using multiple data block replicas to achieve redundancy and reliable, data on different storage nodes have more than one block copy, any node failure, the system will automatically copy the data block copy to the new storage node, the data is not lost, the implementation of data integrity and reliability, and to support the way through the codec algorithm to achieve high reliability, Any damage to multiple storage nodes at the same time, the data can be automatically recovered by decoding algorithm. This feature can be applied to a very high level of data security, at the same time, compared to the reliability of replica redundancy to achieve a significant increase in disk space utilization, less than 40% of disk redundancy can be achieved at any time to damage three storage nodes without losing data.

The Metadata management node uses the high availability of multiple-machine synchronous hot backup to fault tolerance, any server failure, seamless automatic switching to another server, uninterrupted service. The whole system has no single point of failure, automatic hardware fault shielding.

Multi-server Simultaneous external service

SDFs performance, support high concurrency, bandwidth saturation utilization. The cloud storage system separates the control flow from the data stream, while the data is accessed by multiple storage servers and provides high concurrent access. The load is automatically balanced to equalize the access load of different clients to different storage servers. The system performance increases linearly with the increase of node size. The larger the scale of the system, the more obvious the advantages of the cloud storage system and no performance bottleneck. It is advantageous to the formation of super large-scale city-level monitoring network.

Reduce construction and operating costs

Reduce construction and operating costs, as the foundation layer of the entire platform, file systems and data warehouses with Low-cost high storage PC or single board to replace the previous small machines, combined to provide very high computing power and very low storage costs! The use of distributed computing and storage, combined with optimized hardware, through the cluster operation and maintenance management system to achieve computing, storage, network and other resources dynamic distribution and deployment, greatly improve the utilization rate of resources.

Cloud storage will drive the security video surveillance system toward a unified platform for communication, computing, storage and application, rather than a redundant it subsystem in the overall system.

Dynamically increase or decrease storage nodes

All management of SDFs (Skydistributedfilesystem) cloud storage Systems is done by the cloud storage System Management Monitoring Center, and users can easily manage the entire system without any expertise. Through the professional distributed cluster monitoring subsystem to SDFS all nodes to monitor, the user through the interface can clearly understand the operation of each node.

Online scaling, you can, without stopping the service, dynamic addition of new storage nodes, only simple operation, can realize the system capacity from TB level to PB smooth expansion, can also pick up any node, the system automatically reduces the size without losing data, and automatically the picked up the data on the node to other nodes, Ensure that the redundancy of the whole system data is unchanged. and improve read-write and network performance synchronously.

The idea of monitoring storage technology

Intelligent monitoring technology has been in a number of independent areas of development, but there is no more perfect comprehensive solution, the author based on years of work practice, combined with the current development of computer technology, put forward the following five technical aspects of the idea.

Intelligent Code Flow Technology

Intelligent code flow, is the system according to the image recognition, according to the screen movement of the main movement speed, the normal recorded video to adjust the code flow, the final video storage. For the movement of slow motion, the stream record can even be as low as 8 frames/sec (fps) below, for normal speed movement of the main body, code flow set normal 25fps, for sensitive images of moving objects video stream can be set above 30fps. For high-speed movement of the main body, in conjunction with high-speed camera, the code flow can be up to 1000fps or more. The Intelligent Code flow technology can reduce the video storage resources of the insensitive image, thus providing the abundant recording ability for the sensitive image.

Dynamic Region Adaptive Intelligent Monitoring technology

Take a garage video surveillance screen For example, the screen of sensitive information for the movement of vehicles and people. The insensitive picture is the background (ground and roof). But in the actual video recording, the background (ground and roof) occupies more than 50% of the storage space. The image recognition technology can be used to determine the fixed background and moving object image, so it is possible to have the motion only in the recorded picture. Dynamic area Intelligent Monitoring, is only recorded in addition to the background of the moving objects, thus greatly reducing the video storage space requirements, the same storage space, the length of video can be retained can be doubled. Camera system, through the image recognition technology, can learn intelligently, judge which image is the background, even if it is controlled by the PTZ camera, through the set time of automatic learning, can also automatically determine the recording of a new background in the video, so as to achieve the intelligent monitoring of moving object images.

Low-bit-stream recording technology in non-sensitive regions

25% to 30% of the area above the video, usually the top of the sky or building, is basically the area where the sensitive elements are unlikely to occur. For example, some typical monitoring images of red mosaic part of the image, the basic will not contain people care about the video content, so you can put the video image of the insensitive area of the ignore or with a low stream of other records, only in the playback time and high code streaming video to do a synchronization. It is worth noting that the insensitive areas in different scenarios, there are different, and some may be under the video image, some monitoring scenes, insensitive areas are irregular, can be installed in the video monitoring system, according to the actual situation of the application level of artificial settings.

Face/vehicle recognition (or other sensitive moving objects) drives HD video recording technology

In special scenes, such as the entrance to the building, elevators and other places, the face is a sensitive image. In the garage, the entrance and exit of the district, the vehicle and its plate are sensitive images. If all with High-definition video is able to meet the monitoring needs, but video storage, especially for a long time to save will require a large amount of storage space, if the image recognition technology to determine the presence of sensitive images, only to drive the camera to start High-definition records, for general insensitive images, The SD or even the low bit rate video stream is started to record. This combination of HD and SD monitoring records, that is to ensure that the quality of the recording of sensitive images, but also to a large extent reduce the amount of video storage.

Distributed storage technology of video file in sequence frame

The frames generated in one second are identified as sequence frames, the storage and playback sequences are compiled, and the frames of different sequences are divided into several file stores; a single frame sequence file can be played separately, and the effect is equivalent to the video effect of the low stream record.

All frame sequences can be synthesized with full video playback, then high-definition (or SD) video effects. When you need to reclaim storage space, you can plan by storage strategy to cover a portion of the sequence frame video files. The other part is saved to make more efficient use of storage space. For example, according to the original storage capacity can be retained for 1 months of video data, after the distribution of video frame files stored, you can retain a few months of the selected sequence frame video file data. To have retained the medium and long time frame video file partial coverage, realization fades the video gradually discard, the long-term retention video data is not completely disappears, but slowly disappears, discards. This will maximize the time spent monitoring video retention.

Video streaming of intelligent video surveillance based on the above five technologies

The Intelligent monitoring video stream is recorded and stored after a series of different logical layers of image recognition and intelligent judgment. The application of dynamic region Adaptive intelligent monitoring technology and insensitive low bit stream recording technology can be carried through the whole stored procedure according to the actual needs, or selectively implemented according to the manual setting of the application level. Image recognition from the initial identification, to the fine recognition, according to the needs of layered. To the video image recognition sampling frequency, may according to the need, sets the specific frequency, does not need to each frame image recognition.