Huawei's high-definition intelligent video surveillance solution helps efficient highway operations

Introduction: At present, security technology is moving towards networking, high definition and intelligence. Highway video surveillance systems are no exception. However, highway monitoring has the characteristics of complex monitoring subsystems, long distances between camera locations, complex outdoor and tunnel environments, and high vehicle speeds, which puts higher requirements on the stability of the entire monitoring system and the flexibility of application. Huawei uses software-defined cameras and advanced cloud monitoring products to help build a new generation of high-speed monitoring systems for Taihang Mountain, Yanchong, Gansu Weiwu Expressway, and Urumqi to Yuli Expressway.

As early as 2014, my country's expressway mileage reached 100,000 kilometers, surpassing the United States' 92,000 kilometers and becoming the world's first. However, the speed of China's expressway construction has not slowed down. By the end of 2017, my country's expressway mileage reached 135,000 kilometers, accounting for half of the world's expressway mileage. The mileage of a highway is generally more than 100 kilometers, and 300-600 cameras are deployed to monitor various places such as toll stations, tunnels, and outdoor areas. Highway monitoring has the characteristics of complex monitoring subsystems, long distances between camera locations, complex outdoor and tunnel environments, and high vehicle speeds, which poses three major challenges to the deployment of monitoring systems:

1. Monitoring systems require intelligent management

The highway monitoring system can be divided into multiple subsystems such as toll station monitoring, tunnel monitoring, field monitoring, and service area monitoring. Each subsystem has different requirements for cameras. For example, toll stations mainly capture license plates with short focal lengths, while field monitoring mostly uses pan-tilt cameras with long focal lengths (located at long distances), and service areas have greater requirements for coverage. Each subsystem also has different methods and physical locations for video storage. If only manpower is used to manage and operate monitoring equipment, the workload will be huge. Currently, highway monitoring systems are in urgent need of intelligent management.

2. The outdoor environment and the tunnel environment are complex

Highways are fast channels connecting urban areas. The terrain they cross is very complex and the environment is diverse. Haze, rain and snow will seriously affect the monitoring effect of front-end cameras. At present, highways are in urgent need of 24-hour all-weather monitoring in various scenarios.

3. Intelligent analysis and recognition of high-speed vehicle license plates

The high speed on the highway makes it difficult to apply intelligent analysis, which has always been a problem in the industry. The speed of vehicles on the highway generally exceeds 120 kilometers per hour, and even reaches 200 kilometers per hour. At high speeds, the existing side-mounted PTZ cameras cannot accurately recognize license plates due to software and hardware problems such as angles, image sensors and algorithms.

Based on the existing application systems and actual needs of high-speed surveillance system construction, Huawei provides software-defined cameras and advanced cloud surveillance products. In this solution, Huawei has designed an end-to-end solution architecture, adopted open standards and advanced intelligent technologies, and provided targeted professional intelligent security solutions that meet the needs.

Video quality diagnosis + automatic inspection guarantee, helping efficient operation and maintenance

Huawei's intelligent network cameras use an automatic inspection design to poll and detect the device system operation status (temperature abnormalities, SD card read and write abnormalities, etc.) in real time, improving maintenance efficiency and reducing equipment maintenance costs.

1. Camera built-in video quality diagnosis

Using the video quality diagnosis function, the camera automatically analyzes the collected images, extracts features and events to determine changes in the target's morphology, and determines whether the image is abnormal by setting certain conditions and rules. It combines image processing technology, computer vision technology, computer graphics, image analysis and other technologies. It can accurately analyze, judge and automatically alarm common camera faults such as noise, scrolling, blur, color cast, image freezing, gain imbalance, video signal loss, as well as video signal interference, video quality degradation and other problems in the video image.

2. Automatic temperature regulation

Huawei network ball camera supports automatic temperature adjustment function. When the internal temperature of the network ball camera is higher or lower than the safe operating temperature, the internal logic circuit will automatically turn on the ball camera fan or heater for automatic temperature adjustment. Once the temperature is too high to adjust, the camera will automatically generate an alarm signal and upload the alarm information to the monitoring center, waiting for maintenance personnel to perform maintenance.

Scene adaptation to cope with complex and changing environments

High-speed cameras are mostly outdoors or inside or outside tunnels, and the imaging environment is complex. For example, for special weather such as heavy haze and fog in Hebei, the fog penetration algorithm is adapted; for outdoor strong winds in Gansu and Xinjiang, the anti-shake function is adapted; for tunnel scenes, the super wide dynamic function is adapted. However, the environment is constantly changing. Can the camera automatically adapt the imaging parameters according to the environment?

Huawei's X series cameras have a built-in high-performance NPU chip and use a deep learning algorithm. The scene adaptation can automatically identify the dynamic range of scenes by identifying subdivided scenes such as rain, snow, fog, low light, strong light, and high speed, ensuring clear monitoring image quality in all scenes.

Deploy a variety of intelligent analysis systems to help upgrade operations

Traditional vehicle capture cameras use image sensors that scan line by line. When a vehicle exceeds 80 kilometers per hour, the image will have a ghosting phenomenon, and even the best algorithm cannot be recognized. On highways, the speed of vehicles is generally as high as 120 to 200 kilometers per hour. The license plate is the vehicle's ID card and the most critical information. For applications such as vehicle-road coordination, path recognition, and vehicle big data, license plate data is one of the basic data.

To meet the needs of this scenario, Huawei has designed a new generation of high-speed snapshot cameras + self-developed license plate feature algorithms, using a new Pre global shutter CMOS sensor, which changes the line-by-line scanning method to a global scanning method, and supports license plate capture of vehicles traveling at a speed of 200 kilometers per hour. Combined with the self-developed headlight suppression algorithm and de-motion blur algorithm, the license plate recognition accuracy in high-speed scenarios is higher than 95%.

At the same time, the high-performance NPU built into the X-series cameras greatly improves the performance of deep learning algorithms, supports license plate recognition, vehicle model recognition, body color recognition, vehicle logo recognition and vehicle sub-brand recognition, supports illegal detection such as driving against traffic, driving without following directions, speeding/low speed detection, tail number restriction, and supports facial capture of drivers and passengers.

Huawei's IVS intelligent video surveillance system provides multi-dimensional efficiency and convenience for the operations and production of multiple highway customers in terms of efficient management, intelligent operation and maintenance, and service upgrades.

The 21st China Highway Informatization Seminar will be held at Xiamen International Conference and Exhibition Center from March 28 to 29, 2019. Huawei will present advanced and reliable smart highway innovations through keynote speeches, the Vehicle-Road Collaboration Technology Summit, and a large booth built around "Smart Highway". As a leading global provider of ICT (information and communication) infrastructure and smart terminals, Huawei is committed to digital technology innovation in the transportation industry, building an open and cooperative industry ecosystem with partners, jointly promoting the digital transformation of the transportation industry, and becoming a smart transportation engine to connect the world.