Application of LoongShine Panoramic Video System in Expressway Monitoring

Project requirements

In highway monitoring, it is usually necessary to continuously monitor specific sections of hundreds of meters. The traditional way is to configure a camera in each lane to solve the problem of insufficient wide angle of a single camera; In terms of monitoring distance, the method of long focus and fixed focus is often used. The fixed focus can only be clearly displayed in the visible fixed distance range of this focal length, but it can not be clearly identified outside the distance of this focal length, which will lead to many blind areas for monitoring and law-enforcement without evidence.

The use of video relay can significantly improve the problem of blind spots, but it also brings other problems:

1. Infrastructure implementation issues: deploy more focal cameras at a fixed point; Or more fixed frames need to be built within a certain distance, which needs to be coordinated with multiple government departments, and may affect the city appearance and public psychology;
2. The problem of monitoring large screen: it is necessary to broadcast the video of several adjacent cameras on multiple screens or in multiple windows on one monitor, and a moving target will appear on multiple screens or windows, resulting in disorderly display, affecting the interpretation efficiency of staff, and also not conducive to leading video command and decision-making;

In the traditional network scheme, almost all of them have to build a few kilometers of wired network before accessing the private network. The system access is a single route, which not only has a single point of failure, but also makes the system implementation cycle very long and the cost is very high; When the data center fails, the system has no remote disaster tolerance measures, resulting in the system monitoring stop; The monitoring data in the data center has no tamper-proof function, and is prone to malignant events of video evidence loss.

Panorama camera scheme

Intelligent panoramic imaging technology is to integrate multiple photoelectric imaging systems and cooperate with multi-channel electronic processing systems to achieve imaging of hundreds of millions or billions of pixels. This technology can solve the contradiction between the large field of view and high resolution of the traditional photoelectric system, and greatly improve the total amount of information transmitted by the photoelectric system. It integrates multiple focal length cameras in one device, which is easier to deploy, and the panoramic imaging camera supports solar DC power supply.

Through intelligent video splicing technology, 10 million to 1 billion pixels of video images can be output. It can achieve 30 fps real-time splicing at 10 million pixels and 5 fps real-time splicing at 500 million pixels. Through continuous optimization and upgrading of splicing algorithm and GPU hardware, this performance index is still improving.

Panoramic imaging cameras are spliced and fused with cameras of different focus segments to eliminate the monitoring blind area; Using the real-time fusion technology of 14 groups of multi-focus camera video of 150 mm combination, a smooth picture can be obtained in the depth of 3~2000 meters. The effective pixels are 80 million, the camera height is 6 meters, the camera horizontal inclination is 1.5 °, and the face recognition of 200 meters, the license plate recognition of 400 meters, and the human type recognition of 1500 meters can be realized in the sunlight, which can greatly save the capital construction cost. Under the same effect, the original 200 meters of a node will be changed into 2 kilometers of a node.

The virtual camera technology is used to provide external services. The virtual camera can be accessed through HTTP protocol, ONVIF protocol, RTSP protocol, GB-28181 standard and other methods, without changing the topology of the user's existing monitoring platform, and seamlessly access the existing monitoring platform based on the existing network; It supports H264 and H265 video coding formats and can seamlessly connect with third-party video structured systems and coding systems.

The splicing screen of each large monitoring screen corresponds to a virtual camera. The system will cut the splicing frame image into multiple independent virtual camera video frame images according to the number and resolution of the splicing screen, so that each splicing screen can display independently. If the splicing screen is thin, the display effect will be better.

5G transmission scheme

The output bandwidth of the 80 megapixel panoramic camera is about 64Mbps. The adoption of 5G network communication can significantly reduce the cost of infrastructure, and can make it possible to deploy monitoring points that cannot be wired. LoongShine LeoNet 5G router provides a faster, more efficient and lower cost interconnection deployment scheme for panoramic video monitoring, which can realize multiple access aggregation. The system supports two-way NAT, VPN and solar DC power supply. (Note: network bandwidth depends on local network)

Compared with wired network, 5G network is easier to realize multi-route redundancy. Configure multiple LeoNet 5G routers for each panoramic camera. Each 5G router accesses the same or different operators. The 5G router can be configured as an aggregation or active standby mode.

LeoStor distributed storage solution

The panoramic camera needs low-latency NAS read and write, and LoongShine LeoStor scales the NAS storage system horizontally. It is derived from high-performance computing, has advanced distributed storage architecture and parallel storage technology. The NAS read and write under the pressure of 10GB/s can still maintain millisecond latency, perfect data security and redundancy technology, and supports multiple copies and N+M: M erasure code, as well as WORM security technology.

LeoStor supports S3 object storage. The panoramic imaging system can connect multiple data centers through the LeoStor S3 GTM. When a data center fails, the panoramic imaging system polls an available data center. At the same time, the data source of the monitoring system is also allocated to the data center to ensure that the whole monitoring process has a lower flash time of seconds.

The LeoStor distributed storage system can run on the 2U, 3U and 4U x86 servers of Great Wall, Wuzhou, Ultramicro, Huawei, Inspur, Dawn, Lenovo and other companies, as well as the Feiteng ARM servers of Great Wall, Lenovo and the Kunpeng ARM servers of Huawei.

Summary of system scheme

LoongShine panoramic video system includes six subsystems, providing a complete "delivery" system. The panoramic camera and 5G router support solar DC power supply, provide IP67 protection level and link redundancy function, and the operating temperature is - 40 ℃~80 ℃ to meet the deployment requirements of "field"; The splicing video can be cut in real time, which can be displayed on the large screen through the virtual camera. At the same time, the requirements of different resolutions can be met through the transcoding system.

LeoStor modular cabinets only need to be deployed in ordinary clean offices, without professional computer rooms. LeoStor high redundancy architecture, automatic fault recovery, strong self-sustainability, and significantly reduce the cost of operation and maintenance. At the same time, when a data center fails, through the scheduling of LeoStor GTM, it can automatically switch to the standby data center to meet the system-level disaster tolerance requirements of two places and three centers.

Sample piece appreciation

Use the longitudinal multi-focus panoramic camera to take a random shot outside the window, and capture a frame in the high-resolution video. Have you found that the near tens of meters and the far kilometers are clear, which can be used in the scene with obvious longitudinal path, such as urban road sections, roads, railways, marathon and so on; If the horizontal camera module is added, the area of 2000 meters * 2000 meters can be observed, which is the best solution for scenes such as the big square and the performance scene.