Why is the camera restricting the intelligent bottleneck of traffic monitoring?

In urban traffic monitoring, whether it is electronic eye, Skynet monitoring, illegal capture, traffic monitoring, etc., high quality security cameras are inseparable. As the front-end product of the entire traffic video surveillance system, the surveillance camera has always been considered as one of the core parts of the traffic monitoring system. From a certain perspective, the intelligentization of the camera has promoted the development of urban intelligent traffic monitoring systems. In this process, the Internet of Things technology plays a catalytic role.

This article will analyze how the "bottleneck" breaks through the characteristics of the surveillance camera and its historical process.

First, intelligent cameras can meet the current traffic monitoring needs

First of all, it is necessary to clarify the specific characteristics of traffic monitoring. Traffic monitoring is mainly composed of two parts: the traffic information statistical system and the electronic police punishment system. There are two main features:

1. Real-time monitoring of road conditions: The camera at the traffic intersection transmits the real-time image of the monitored area back to the traffic command center, so that the traffic commander can grasp the traffic flow of the road sections, whether the traffic lights are working normally, whether there are violations and traffic accidents, etc. .

2, real-time video: video surveillance images will be saved to the video command center of the traffic command center, as a basis for dealing with traffic accidents, violations, and even public security crimes and other emergency conditions.

In order to achieve a clear picture, and to see the hard-wearing requirements such as license plates at night, the camera needs to have multiple functions and good performance.

First of all, the camera must have long-term stable operation in various harsh environments to ensure that the image quality is not deteriorated and the failure rate is low. If the "eyes" of a traffic monitoring system are not good or often go wrong, even if the rest of the system is perfect, the defects of the whole system are clear.

Second, the camera is highly adaptable to changes in light. The luminosity difference between day and night is quite large, and the difference between the scorching sun and the rainy days;

Third, the camera should be able to capture and shoot at high speed.

In order to meet the current high-quality traffic monitoring needs, traffic surveillance cameras have been promoted from traditional to networked, high-definition, intelligent and Internet of Things.

Second, the progress of the camera promotes the intelligent process of traffic monitoring

At the end of 1868, the world's first traffic signal appeared in London, England, a glass signal light that burned gas. However, the good weather did not last long. The first traffic light that only ran for 23 days suddenly exploded and destroyed a police officer. The British government immediately stopped using the signal.

But this 23-day-old signal light ignited the passion of developing traffic lights throughout Europe and the whole world. Over the next hundred years, various traffic lights appeared.

In the early days, when vehicles and pedestrians passed through various intersections with traffic lights, they were mainly dispatched by traffic police. However, with the increase of vehicles and the development of urban roads, the limited traffic police can no longer meet the command and violation monitoring requirements of traffic signal intersections, and traffic surveillance cameras have emerged.

In 1991, the world's first surveillance camera "CoffeeCam" appeared at the Teja House Coffee Shop at the University of Cambridge, England. Since then, surveillance cameras have been gradually applied in the field of traffic monitoring such as electronic eye, Skynet monitoring, illegal capture, and traffic monitoring.

China's traffic monitoring has experienced six stages, and these six stages can be said to be caused by the performance changes of the camera.

Intelligent Transformation of Traffic Monitoring in China

The first stage (before the 21st century): digital camera + ground coil

Since the early camera was very backward, traffic monitoring was initially implemented as a “digital camera + ground-sensing coil”. Although it was able to capture illegal vehicles, this method mainly used civilian digital cameras for capturing, which was not stable.

Secondly, due to the rapid replacement of digital cameras, the same type of digital camera is difficult to find after one or two years, causing trouble for subsequent repair and replacement; and the digital camera can not achieve continuous capture, the system can not achieve the bayonet function.

In addition, this method uses a ground-sensing coil for triggering, and it is necessary to lay a ground-sensing coil at the front end, and the ground-sensing coil needs to cut a circular line groove in the lane, and then embed the induction coil. When the vehicle passes, the induction coil will send a signal to the corresponding equipment, and the detection is relatively accurate. However, it is necessary to embed an induction coil in the ground, which increases the difficulty of construction and increases the cost of the project. Moreover, it is troublesome to re-embed the coil when the road surface is changed.

In addition, the high-latitude open freezing period and the low-latitude summer pavement, as well as the poor road quality, the coil maintenance work is huge, and the coil needs to be replaced again after a period of time, and the construction and maintenance costs are quite high.

The second stage (2001 - 2005): SD camera + digital camera

This stage is implemented as "standard definition camera + digital camera". After 2000, standard-definition cameras appeared on the market, making it possible to directly detect illegal lights such as red lights and violations by analyzing video. This kind of implementation avoids the shortcomings of the ground sense coil, but still uses the civilian digital camera to capture, which has problems such as poor stability, fast update, and difficult maintenance. In addition, due to technical limitations, early digital cameras could not achieve continuous capture, and there would be some omissions in the single capture mode of illegal activities such as red light.

The third stage (2006 - 2007): SD camera + SD camera

At this stage, the way of traffic monitoring is changed from “SD camera + digital camera” to “SD camera + SD camera” mode. Since the industrial-grade SD camera directly replaces the civilian digital camera, the disadvantages of using a civilian digital camera are avoided, and the performance is relatively stable.

However, due to the low pixel size of the SD camera, the resolution of the picture is up to 768×576. Although it can meet the basic management requirements of vehicle detection, it is still impossible to see the license plate and red light signals at the same time. For the public security department, the auditing of the escaped vehicles and decks The facial features of drivers, such as vehicles and blacklisted vehicles, are even more stretched. And the amount of construction is also large, a lane needs a camera to do close-ups.

The fourth stage (2008 - 2009): HD CMOS camera + ground coil

There are two main ways to implement the electronic police system at this stage: high-definition CMOS camera + ground-sensing coil, standard-definition camera + high-definition CMOS camera. Due to the use of industrial-grade HD CMOS cameras, the performance of the camera is relatively stable. At the same time, the camera can capture continuously to achieve the function of bayonet monitoring. However, due to the high-resolution CMOS camera, the imaging effect is rough (not sensitive to light changes, sometimes catching pictures can not see the traffic lights, affecting the punishment afterwards), and the frame rate is low (generally 2~3 frames/second, it is impossible to realize video analysis vehicle behavior and video recording). The shortcomings of the function) still do not achieve the best results.

The fifth stage (2010 - 2015): HD CCD camera

With the popularization of large frame rate HD CCD cameras and the continuous optimization of video analysis algorithms, the traffic monitoring at this stage is mainly implemented by “HD industrial grade CCD camera”. Only a part of the image of the license plate of a single lane can be captured, and the standard-defining equipment that cannot balance the characteristics of the vehicle and the driver's face will gradually fade out of the historical stage and replace it with a high-definition device.

Sixth stage (2016-present): Smart camera monitoring

At this stage, it mainly benefited from the development of Internet of Things technology, computer vision technology, intelligent analysis technology and digital image processing technology in recent years.

IoT technology means that current urban road monitoring is no longer limited to video surveillance. Combined with Internet of Things technology, it can quickly publish road condition information and road congestion status to the information board, so that drivers can choose favorable routes.

Not only that, the application of Internet of Things technology enables the urban road monitoring system to be connected to various IoT sensors of the road test, or to the sensors of the vehicle itself, such as weather stations, parking lots, bus station information display screens, GPS modules of vehicles. In addition, various road conditions such as rain and snow can be displayed in a three-dimensional display on the urban road monitoring system, thereby facilitating the rapid response of the traffic control department and reducing the losses caused by various extreme weather conditions.

Computer vision technology processes, analyzes and understands video signals. By automatically analyzing sequence images without manual intervention, it locates, identifies and tracks changes in the monitoring scene, and analyzes and judges them on this basis. The behavior of the target.

It is understood that the current intelligent camera monitoring in China can effectively complete the automatic monitoring of vehicle deceleration, vehicle retrograde, traffic jam, road smoke and fire, and predict and analyze the traffic flow, vehicle speed, vehicle type, and emergency urgency. Road safety operations and hazardous situation rescue provide the necessary data support.

Intelligent video analysis and behavior recognition technology is based on video image feature learning, target object detection and behavior recognition, long-distance trajectory tracking and other algorithms and techniques to analyze the video stream of the monitored vehicle to extract and identify the target object's violation Behavior; simultaneous intelligent video analysis technology can improve the accuracy of monitoring.

4G mobile communication technology video surveillance terminal transmits captured video images to the background server for viewing and storage through 4G network. The application of 4G technology reduces the cost and complexity of wired communication network deployment, and the system construction cycle is no longer subject to site and construction wiring. The restrictions thus save a lot of labor and capital investment.

to sum up:

The intelligent process of the camera not only solves the demanding requirements of climate, light, duration, and accuracy of traffic monitoring, but also spawns a billion-dollar intelligent security market including the Tianye system and home security. In addition to the players in the traditional security field, Internet giants have also entered the game.