Wrong way driving detected in the camera and automaticly alerts the TMC
Intelligent traffic cameras open up a host of possibilities for traffic planners and controllers alike.
If traffic management centres (TMCs) around the world are to cope with the increasing demands of growing traffic flows while maintaining or improving transport safety and efficiency, then video monitoring will have to be supplemented by automated warnings of incidents or deviations.
According to Patrik Anderson, business development director at Swedish camera manufacturer2215 Axis Communications, it is not cost-efficient or necessary to watch live video from all cameras at all times. Instead, based on their experience and on what is currently happening, each operator will actively choose what live streams to monitor and when an incident is reported the relevant cameras can quickly be selected to view the situation.
However, this inevitably means there are cameras that are not being constantly monitored and operators often rely on commuters reporting deviations. The question is, without users reporting a problem, how long would it take to discover an incident that happened in view of a camera that is not being displayed on the video wall?
“Moving forward, intelligent video analysis will play a much more instrumental role in traffic management. Cameras with intelligent capabilities can automatically alert TMC staff to deviations such as congestion, stopped vehicles and wrong-way driving. Instead of waiting for a road user to report an incident or traffic deviation, the traffic management operators will be immediately alerted by intelligent cameras. In addition, the smart traffic cameras can also provide real-time or historic traffic data,” says Anderson.
Axis lays claim to inventing network cameras (often called IP cameras) in 1996 and today compression protocols such as H.264 mean it is technically possible to feed an increasing number of video feeds into TMCs. What hasn’t evolved as rapidly is the human ability to monitor the growth in the number of video streams so there is likley to be ocassions when an incident that happens in full view of a surveillance camera could go unnoticed for a period of time. And any delay in detecting incidents in safety-critical situations such as hard shoulder running, could have significant consequences.
Axis believes it has come up with an answer in the form of its latest generation of network cameras. These include a powerful processor to enable real-time video analysis to be done by the camera itself rather than streaming all video data to the TMC for central video analysis and viewing. These features are added to the company’s 60 frames per second camera which also feature built-in capabilities to manage difficult light situations such as head-light blooming and wet pavement reflections.
Anderson says that Axis has partnered with a number of software companies to develop intelligent traffic applications for Axis’ cameras. The applications enable the cameras to alert the TMC in real time if one of a number of scenarios occur - effectively transforming them from passive onlookers into active sensors. “It also means there is no need to constantly transfer all the video from all cameras all of the time; images only need to fed to the TMC when an incident occurs or if the operator requests a live view,” he says.
This produces a considerable saving on bandwidth and system capacity while the in-camera processing enables other functions to continue even during periods when the camera’s video stream is not being transmitted. For instance a camera with data collection capabilities can be set to only transfer data – for instance the number of vehicles passing – which requires much less bandwidth than complete video data. Yet it can also be used for live video viewing at any time when requested by the operator.
According to Anderson, this gives the system designer the flexibility to choose how much video should be streamed. “Depending on how the TMC wants its network architecture configured, video systems cameras can be set up to only transmit live images when an intelligent application has detected an incident. The system can also be structured so cameras only transmit low resolution images unless an incident occurs, and if it does the frame rate and image quality is automatically increased as well as an alert being sent. Also, if during their normal monitoring the traffic operator manually selects a camera, it will automatically increase the quality to provide a full HDTV quality display.”
As each intelligent camera on a network can be individual programed each can be to set up to detect a different scenario depending on their location (incident detection, ANPR, vehicle counting and so on) and alert the TMC if and when required. Partner applications run in parallel with the camera’s regular operation without interfering or limiting performance, so live video can be transmitted even when the smart application is running. Currently, a camera can run one intelligent partner application in parallel with its normal operation and in the near future Axis says it will be possible to run multiple applications simultaneously.
All applications can be installed and configured remotely over the network using a web browser. Where the traffic authority has various applications in many different cameras, multiple applications downloads and upgrades can be done simultaneously. Anderson says traffic authorities or system integrators can choose applications from a range of suppliers as many leading providers have chosen to develop applications that can run directly on its cameras.
In addition to those listed above, these include traffic flow analytics, vehicle count and classification, speed drop, stationary vehicle and wrong-way drivers. When a camera has detected an incident, the TMC operator is immediately alerted and provided a real-time view of the situation. This will allow TMCs to quickly and efficiently respond to traffic incidents, adapt variable messaging signs and/or red lights and inform road users, first responders, maintenance crews and the media about the current traffic situation.
While the network camera has both the hardware and software to enable it to run video analytics, there is an on-cost for adding an intelligent application to each camera. However, Anderson points out that software costs will be incurred anyway and have to be offset against the savings from not needing large and very powerful servers at the TMC to process all the incoming live video streams.
If cameras are to move from a monitoring role to that of an active sensor and if TMCs are to use the automated information they supply, then reliability becomes a key issue – it is one thing to be unable to view an incident; it’s quite another to not receive an expected alert if such an incident occurs in the first place. This means cameras will have to be able to capture what is occurring at dusk, when it is dark and in bad weather with enough clarity for the processing software to process and analyse those images.
“If you add to this mix, head and tail-lights of vehicles and wet roads that can blind, reflect or bloom out camera sensors, you have a complex task for any camera to handle,” says Anderson before adding that Axis’ cameras contain wide dynamic range technology that can handle bright light and darkness/shadows at the same time.
Such systems must also cope with extreme temperatures and notify the TMC if for some reason it cannot provide the normal processing and alerts. Indeed system reliability and connectivity will have to be assured before TMCs can depend on cameras as sensors in safety-critical situations. Anderson says even if authorities are not planning to implement automated incident detection, when new surveillance systems or upgrades are undertaken, consideration should be given to ensuring the specification provides the resilience and technical scope to do so in the future.
“We believe in the future it will not possible for TMC staff to monitor all the inputs at their disposal, some may say it is already happening. This is a human limitation and not a technical one – and there is a technological remedy.”
If traffic management centres (TMCs) around the world are to cope with the increasing demands of growing traffic flows while maintaining or improving transport safety and efficiency, then video monitoring will have to be supplemented by automated warnings of incidents or deviations.
According to Patrik Anderson, business development director at Swedish camera manufacturer
However, this inevitably means there are cameras that are not being constantly monitored and operators often rely on commuters reporting deviations. The question is, without users reporting a problem, how long would it take to discover an incident that happened in view of a camera that is not being displayed on the video wall?
“Moving forward, intelligent video analysis will play a much more instrumental role in traffic management. Cameras with intelligent capabilities can automatically alert TMC staff to deviations such as congestion, stopped vehicles and wrong-way driving. Instead of waiting for a road user to report an incident or traffic deviation, the traffic management operators will be immediately alerted by intelligent cameras. In addition, the smart traffic cameras can also provide real-time or historic traffic data,” says Anderson.
Axis lays claim to inventing network cameras (often called IP cameras) in 1996 and today compression protocols such as H.264 mean it is technically possible to feed an increasing number of video feeds into TMCs. What hasn’t evolved as rapidly is the human ability to monitor the growth in the number of video streams so there is likley to be ocassions when an incident that happens in full view of a surveillance camera could go unnoticed for a period of time. And any delay in detecting incidents in safety-critical situations such as hard shoulder running, could have significant consequences.
Axis believes it has come up with an answer in the form of its latest generation of network cameras. These include a powerful processor to enable real-time video analysis to be done by the camera itself rather than streaming all video data to the TMC for central video analysis and viewing. These features are added to the company’s 60 frames per second camera which also feature built-in capabilities to manage difficult light situations such as head-light blooming and wet pavement reflections.
Anderson says that Axis has partnered with a number of software companies to develop intelligent traffic applications for Axis’ cameras. The applications enable the cameras to alert the TMC in real time if one of a number of scenarios occur - effectively transforming them from passive onlookers into active sensors. “It also means there is no need to constantly transfer all the video from all cameras all of the time; images only need to fed to the TMC when an incident occurs or if the operator requests a live view,” he says.
This produces a considerable saving on bandwidth and system capacity while the in-camera processing enables other functions to continue even during periods when the camera’s video stream is not being transmitted. For instance a camera with data collection capabilities can be set to only transfer data – for instance the number of vehicles passing – which requires much less bandwidth than complete video data. Yet it can also be used for live video viewing at any time when requested by the operator.
According to Anderson, this gives the system designer the flexibility to choose how much video should be streamed. “Depending on how the TMC wants its network architecture configured, video systems cameras can be set up to only transmit live images when an intelligent application has detected an incident. The system can also be structured so cameras only transmit low resolution images unless an incident occurs, and if it does the frame rate and image quality is automatically increased as well as an alert being sent. Also, if during their normal monitoring the traffic operator manually selects a camera, it will automatically increase the quality to provide a full HDTV quality display.”
As each intelligent camera on a network can be individual programed each can be to set up to detect a different scenario depending on their location (incident detection, ANPR, vehicle counting and so on) and alert the TMC if and when required. Partner applications run in parallel with the camera’s regular operation without interfering or limiting performance, so live video can be transmitted even when the smart application is running. Currently, a camera can run one intelligent partner application in parallel with its normal operation and in the near future Axis says it will be possible to run multiple applications simultaneously.
All applications can be installed and configured remotely over the network using a web browser. Where the traffic authority has various applications in many different cameras, multiple applications downloads and upgrades can be done simultaneously. Anderson says traffic authorities or system integrators can choose applications from a range of suppliers as many leading providers have chosen to develop applications that can run directly on its cameras.
In addition to those listed above, these include traffic flow analytics, vehicle count and classification, speed drop, stationary vehicle and wrong-way drivers. When a camera has detected an incident, the TMC operator is immediately alerted and provided a real-time view of the situation. This will allow TMCs to quickly and efficiently respond to traffic incidents, adapt variable messaging signs and/or red lights and inform road users, first responders, maintenance crews and the media about the current traffic situation.
While the network camera has both the hardware and software to enable it to run video analytics, there is an on-cost for adding an intelligent application to each camera. However, Anderson points out that software costs will be incurred anyway and have to be offset against the savings from not needing large and very powerful servers at the TMC to process all the incoming live video streams.
If cameras are to move from a monitoring role to that of an active sensor and if TMCs are to use the automated information they supply, then reliability becomes a key issue – it is one thing to be unable to view an incident; it’s quite another to not receive an expected alert if such an incident occurs in the first place. This means cameras will have to be able to capture what is occurring at dusk, when it is dark and in bad weather with enough clarity for the processing software to process and analyse those images.
“If you add to this mix, head and tail-lights of vehicles and wet roads that can blind, reflect or bloom out camera sensors, you have a complex task for any camera to handle,” says Anderson before adding that Axis’ cameras contain wide dynamic range technology that can handle bright light and darkness/shadows at the same time.
Such systems must also cope with extreme temperatures and notify the TMC if for some reason it cannot provide the normal processing and alerts. Indeed system reliability and connectivity will have to be assured before TMCs can depend on cameras as sensors in safety-critical situations. Anderson says even if authorities are not planning to implement automated incident detection, when new surveillance systems or upgrades are undertaken, consideration should be given to ensuring the specification provides the resilience and technical scope to do so in the future.
“We believe in the future it will not possible for TMC staff to monitor all the inputs at their disposal, some may say it is already happening. This is a human limitation and not a technical one – and there is a technological remedy.”
