OMRON

Global

  1. Home>
  2. Research & Development>
  3. Technology in Focus>
  4. Vision Sensing>
  5. Keeping Two Eyes Out for Traffic Conditions

Keeping Two Eyes Out for Traffic Conditions

  • Text Size
  • Small
  • Medium
  • Large

In 1990, the first car navigation systems were introduced in Japan. In just over 10 years, the capabilities of these systems have advanced to an incredibly high level. One of the most useful functions of a car navigation system is to provide vehicles with information regarding traffic conditions. OMRON commands a large share of the market for sensors that are capable of detecting this information and measuring congestion. Taking advantage of its Silhouette Vision technology, OMRON continues to produce developments that enable car navigation systems to offer more accurate information on a real-time basis.

Sensors-the secret behind the navigation system's ability to monitor traffic in various conditions.

A car navigation system is a handy tool for avoiding adverse traffic conditions or for determining a driving route for out-of-town trips. The key components that enable navigation systems to provide traffic information on a real-time basisare sensors. These devices can detect the number of vehicles currently on the road and their driving speed, as well as accidents and other abnormal road conditions, and transmit this information to the traffic control center for processing.

Mechanism behind providing road information
Mechanism behind providing road information

Among the sensors currently on the market, OMRON's stereo traffic flow sensor using Silhouette Vision technology is earning an especially good reputation among users. Known as the Silhouette Vision Sensor (SVS), it is the only sensor in the world that uses two cameras. This two-camera configuration enables the SVS to maintain a recognition accuracy of over 97% in virtually all conditions.

Example of Silhouette Vision Sensor installation
Example of Silhouette Vision Sensor installation

A sensor capable of distinguishing individual cars in any circumstance.

You may ask, "How does the use of two cameras raise the sensor's recognition accuracy?" As mentioned previously, a side view is most effective for identifying the shape of a car, but it's not always possible to install roadside cameras in a position that provides that perspective. In fact, on the road, cameras are usually installed at positions that look down on vehicles diagonally. This kind of installation often presents situations in which several vehicles overlap each other in an image. For conventional single-camera systems that detect vehicles based on changes in gray scale, this makes it very difficult to distinguish one vehicle from another. With Silhouette Vision technology designed to recognize vehicles by means of their side-view silhouettes, OMRON's sensor can capture a stereo image of vehicles using two cameras and convert it into a side-view or top-view shape, thus enabling overlapped vehicles to be distinguished easily.

Conversion of a stereo image into a side-view silhouette using two cameras
Conversion of a stereo image into a side-view silhouette using two cameras

Overlapping is not the only problem confronting conventional sensors. Others include an inability to recognize dark-colored cars at night or during twilight hours. Sometimes these systems incorrectly identify a shadow as a vehicle. Silhouette Vision technology resolves all of these problems with its ability to capture three-dimensional information.

Silhouette Vision Sensor can effectively cope with adverse conditions that make detection difficult for conventional image sensors.
Silhouette Vision Sensor can effectively cope with adverse conditions that make detection difficult for conventional image sensors.

Silhouette Vision has the potential to detect not only the flow of vehicles but that of people as well.

Development is currently underway at OMRON to apply Silhouette Vision technology not only to vehicle detection but also for sensing the flow of people. By capturing the shape of a person three-dimensionally, it should be possible to distinguish a human being from non-humans, and detect the person's motion as well. But in what ways could this be useful? One possible application is counting the number of people that pass through or assemble in a given location. By installing cameras at the entrance of a department store or theme park, for instance, the Silhouette Vision Sensor can count incoming and outgoing people at the same time, providing a precise assessment of the number of customers. Also, with the ability to count the number of people within a certain area, the sensor can be used for counting the number of people in an elevator or the number of people standing in a line, for instance. For privacy protection, Silhouette Vision can also be configured without camera image output.

Incoming and outgoing customers are counted separately to check the number of people currently inside the building.
Incoming and outgoing customers are counted separately to check the number of people currently inside the building.

Another exciting possibility is the analysis of crowd flow. This would be especially useful for convenience stores and supermarkets because it makes it possible to observe the ability of each sales counter to attract customers. Analyzing the flow of customers in a store can als be a marketing tool for stores by suggesting the most effective store layout.The Silhouette Vision Sensor can also be used to measure how long a person stays within the camera's photographic area. This function is useful for creating a security system capable of detecting the presence of a suspicious person.For instance, a person who approaches a building but hesitates to go inside, lingering in a suspicious manner.

Top of Page