Increasing rate of traffic injuries highlights the need for adopting appropriate measures so that automotive travel becomes a safer and more practical way of transportation. In addition to stringent implementation of traffic rules and infrastructure improvements, automotive technology should play an important role in the development of innovative concepts to ensure improved traffic safety. To this end, high-resolution image sensors with high dynamic range and sensitivity will considerably contribute to this goal.
Measures for Road Traffic Safety
The measures needed to improve traffic safety include three issues and they are known as 3E's of road safety: Engineering, Education and Enforcement.
Engineering, the more technical aspect of the 3E's, has advanced tremendously over the last 20 years. This can be seen by way of better traffic signs, improvement of road intersections, installing speed bumps in front of schools and day-care centers, and alerting drivers via vibration inducing striping. All these aspects have significantly contributed towards improving road traffic conditions. Advanced automotive technologies, like ABS, airbags, seatbelts, child seats, etc. are aimed to reduce traffic casualties. Soon, these technologies will be complemented by intelligent traffic infrastructure with real-time traffic flow control.
Traffic education and driver training are other important ways of preventing traffic accidents and reducing the number of casualties. A comprehensive analysis of traffic accidents and examining the most prevalent causes of accidents could help in improving traffic safety.
Traffic Rule Enforcement
Enforcement of traffic rules is confined to detecting red light violations, inappropriate speed, lack of wearing helmets and seatbelts, using phones or data devices while driving, and driving under the influence of drugs or alcohol. This is where high resolution traffic cameras would contribute immensely.
Traffic Surveillance Cameras
Under difficult lighting conditions or fluctuating weather conditions, capturing clear and vivid images of fast moving cars places significant demands on the dynamic range, resolution, and image sensors. So far, image sensors selected from the industrial product lines were somehow able to adapt to the needs of traffic camera applications. However, this does not work all the time.
To overcome this issue, CMOSIS and GATSO are using a different approach. CMV20000, the 20-megapixel CMOS image sensor, was specifically developed for road traffic surveillance, but also proved suitable for medical electronics, industrial imaging, and automation processes.
CMV20000: A Versatile Image Sensor
CMV20000 for traffic surveillance tasks.
Built in CMOS technology, the CMV20000 image sensor provides a pixel grid of 6.4 p., a resolution of 5120 x 3840 pixels, and an active sensor area of 32.8 x 24.6 mm. This matches the optical format of 35 mm cameras. The image sensor features a global shutter that operates in pipelined fashion. Using the combination of correlated double sampling and CMOSIS' 8-transistor sensor cell, dark noise is significantly reduced. Moreover, non-uniformities of the sensor matrix are also reduced, thus making the sensor to cover a wide dynamic range of 66 dB. The sensor provides data in a 12bit digital format through 16 LVDS outputs, at a clock rate of 480 MHz and a frame rate of 30 full images per second. Users can utilize the sensor as an independent subsystem for image capturing.
Programming of CMV20000 Image Sensor
The relevant parameters of the CMV20000 sensor can be easily programmed through a serial interface, making it adaptable to a variety of applications. For instance, exposure time can be set internally which can also be controlled by externally applied signal. This allows users to make a basic adjustment to the existing illumination levels of the scenes and objects to be captured.
Sensor's sensitivity reduces with increasing illumination levels
Poor illumination at traffic scenes is a common scenario, license plates of cars and other vehicles are often partly shaded by the vehicles’ body, making it more difficult to identify them. In order to accommodate these differences in different illumination levels, the CMV20000 provides a unique method for programming segmented conversion characteristics. This extends the dynamic range of the sensor to over 90 dB. In addition, the sensitivity of the sensor reduces considerably with increasing illumination levels.
Flexible Windowing and Subsampling Functions
In majority of traffic applications, the interesting image areas are the ones that show the car license plates of traffic rule violators. The sensor array features a versatile windowing function to allow these partial views. Approximately eight line-wide windows can be programmed and read out at a higher frame rate. This feature can also reduce the data volume to be processed. In addition, the CMV20000 includes a subsampling function for quick capture of images and produces a preview image consisting of a few lines. In colored sensor, this amounts to line pairs to eliminate fake color renderings of the Bayer pattern. This pattern is utilized to train the camera’s LVDS interface for appropriate timing of the output words.
Wide Data Stream
In order to accommodate the vast data volume of the captured images, 16 LVDS output channels are provided that offer a bandwidth of 480 Mbit/s. In case these data rates are very high for a particular application, the output can be reduced down to eight output channels and this will deliver half the frame rate. Also, power consumption can be reduced by switching off the non-used output channels.
CMOSIS CMV20000: Traffic Surveillance Camera
CMOSIS has specifically designed the CMV20000 image sensor for the GT20 traffic camera developed by GATSO. The camera equipped with this sensor sets a new benchmark for efficient traffic surveillance systems. It includes four Intel Atom processors, which offer sufficient computing power to capture red-light violators and to automatically identify car license plate at a frame rate of 30 frames per second. This configuration also helps in recording multiple and repeated traffic rule infringements.
Traffic surveillance unit
The GT20 traffic surveillance camera, equipped with high-resolution CMV20000 image sensor, can be used in challenging applications. The technology overcomes the limitations of current systems and surpasses the scope of traditional speeding and red-light violation detection tasks.
CMOSIS is a supplier of advanced off the shelf, customised and full custom CMOS image sensors developed by multi disciplinary team of image sensor experts. They offers CMOS image sensor products optimized for specific application fields. These products feature high-performance pixel architectures, high frame rates, on-chip ADC, high dynamic range operation modes and high-speed digital interfaces. They are used in 3D imaging, automotive, motion anaysis, scientific and security fields.
This information has been sourced, reviewed and adapted from materials provided by CMOSIS.