Traffic and Transportation (ITS)

Overview

Applications

License plate recognition

It’s necessary to identify license plate numbers on vehicles for various reasons such as identifying cars travelling above the speed limit on highways, managing activity at automatic toll gates, and managing car parks. It is important to select the image sensor best suited to each specific task and environment in which these images will be captured.

When capturing images of fast-moving vehicles with or without flash lighting

Global shutter image sensors facilitate imagery with no distortion when capturing vehicles traveling at high speeds. Where flash lighting is required for night-time imagery or in otherwise dark locations, global shutter image sensors enable a shortened duration of flash lighting. This helps to reduce power consumption compared to imaging systems that use rolling shutter image sensors. In turn, this leads to a more energy-efficient system with lower running costs.

Impact on images when shutter type and flash duration vary

Global shutter image sensors take pixel information from all pixels at once, while rolling shutter image sensors obtain images through line-sequential scanning. This difference results in a longer flash duration to achieve the same exposure time for the rolling shutter type compared with the global shutter type. If the flash duration is not long enough to cover the exposure time, the resulting image will show some parts not illuminated, creating an irregular contrast known as the flash band.

Therefore, global shutter image sensors are suitable for capturing fast-moving vehicles (at speeds over approx. 120 km/h) and/or for reducing flash duration.

When capturing images of vehicles traveling at low speeds

Rolling shutter image sensors, by contrast, are preferred for capturing images of vehicles on ordinary roads (at speeds typically below 120 km/h), as cameras with these image sensors are more sensitive^*1 and less pricey than those with global shutter image sensors.

*1) When comparing images of the same pixel size

- Rolling shutter image sensors suitable for low-speed vehicle imagery: IMX585, IMX664, IMX662

When capturing images in a wide area

The required number of pixels for license plate recognition depends on imaging conditions such as distance to target, vehicle speed, angle of view, and illuminance.

For monitoring a road with several lanes, a multi-pixel image sensor helps minimize the amount of cameras and lighting equipment. This lowers the cost of initial installation and reduces the system footprint (optimizing camera distribution). With a long-term perspective, it also cuts maintenance costs.

Take, for example, the task of monitoring three lanes of traffic simultaneously with a camera mounted on a cross bridge. The required number of pixels is 8M or more, as the camera is far away from the targets and the angle of view is wide.

Patterns of capturing vehicle images on three lanes and comparison of license plate images

When capturing images in a small area

To identify vehicles that pass relatively close to the camera, such as at a car park boom barrier, sufficiently clear imaging can be achieved with a resolution of 2M pixels.

Illustrative setting:
A 2M-pixel camera and flash light set is installed at a car park boom barrier

Captured image sample:
The image is of an actual vehicle at a car park exit, captured under the conditions illustrated on the left. The license plate is clearly captured and the information is legible.

- Rolling shutter image sensors suitable for small-area imagery such as at a car park: IMX662, IMX664

Road monitoring

Capturing outdoor images often means dealing with various adverse conditions including night-time low light conditions and backlit conditions.

On highways, an accurate understanding of events is crucial for making immediate decisions to control the traffic and dispatch emergency vehicles. Whatever’s happening at the time, it’s essential that imaging clearly reveals traffic congestion, weather conditions, obstacles in the road, vehicular accidents, and more.

When capturing images in a low-light environment

The image sensors suitable for capturing images under challenging low-light conditions, such as at night-time, are our high-sensitivity image sensors with STARVIS / STARVIS 2 Technology.

Without STARVIS 2 technology

With STARVIS 2 technology

As shown above, the vehicle shape can be clearly discerned in the image captured using the image sensor with STARVIS 2 technology, despite hardly any roadside lighting and the scene being lit only by the vehicle’s head lights.　

- Image sensors with STARVIS 2 technology, suitable for capturing a wide area in a low-light environment: IMX585, IMX678

When capturing images in a backlit setting

Image sensors with wide dynamic ranges are suitable for capturing images in a backlit setting, where both bright and dark areas are in view. The high dynamic range (HDR) feature enables a clear image to be captured, even of a moving target, without allowing over-exposure.

HDR turned off

HDR turned on

- Image sensors with STARVIS 2, with wide dynamic ranges, suitable for capturing images in backlit contexts: IMX585, IMX678

For more information on achieving high sensitivity and eliminating blowouts, read: STARVIS / STARVIS 2 Technology

Verifying the number of in-vehicle people on HOV lanes^*3

Traffic surveillance cameras can fail to recognize individuals inside a vehicle, depending on camera angles during certain hours of the day, due to light reflection on the front windscreen. This problem can be avoided by properly configured polarization image sensors to reduce or remove this reflected light.

*3) High-occupancy vehicle lanes are restricted lanes for vehicles complying with vehicle occupancy requirements

Image of a vehicle interior taken by a surveillance camera

Standard camera image

Image by a camera with polarization image sensors

While it is possible to apply a polarizing filter to a standard camera to remove the reflected light, the effect is compromised depending on target vehicle models (for example, varying windscreen angles). The polarization image sensor can cancel the effect of reflected light on any vehicle models, improving the accuracy of vehicle occupancy recognition. Watch the video below and review this page for more details.