Sensor technology advances increases ITS opportunities

Basler’s Enzio Schneider explains why advances in CMOS technology provides new opportunities for vision-based ITS applications. Since the beginning of 2015, or even before, it seems obvious that all roads in vision-based ITS applications lead in one technological direction – CMOS. Initially perceived as a trend in vision technology, it has taken a step towards status as the new benchmark with Sony’s announcement to discontinue their CCD production. CMOS sensor technology has become the future for industrial
CMOS sensor
CMOS sensor
Basler’s Enzio Schneider explains why advances in CMOS technology provides new opportunities for vision-based ITS applications.

Since the beginning of 2015, or even before, it seems obvious that all roads in vision-based ITS applications lead in one technological direction – CMOS. Initially perceived as a trend in vision technology, it has taken a step towards status as the new benchmark with 576 Sony’s announcement to discontinue their CCD production. CMOS sensor technology has become the future for industrial and network cameras in intelligent traffic and transportation systems.

Whatever the application, there’s one crucial requirement they all have in common – excellent image quality. Once acknowledged as the traditional strength of CCD sensors, image quality in CMOS sensors has lately taken a major step forward thanks to the significant progress this sensor technology has undergone in recent years. This includes major strides in two important parameters: sensitivity and noise level. As challenging light conditions are an inevitable element of most ITS applications, a sensor’s noise behaviour, sensitivity, and quantum efficiency are inextricably linked to its performance and efficiency in such systems. Modern CMOS sensors such as the IMX174 and IMX249 from Sony’s Pregius series, for example, are technological powerhouses: small in price, but very powerful in performance.

They combine all of the above properties with global shutter and a broad dynamic range, at a very favourable performance/price ratio. With all these benefits combined, they are destined to make it onto any integrator’s shortlist when it comes to selecting potential sensor types for a specific ITS application.

Other major advantages of CMOS sensors in traffic applications, besides their low noise levels, are the absence of smearing and blooming. These artefacts frequently occur in images acquired with CCD sensors.

They compromise the usability of the image material that has been recorded for traffic enforcement and analysis purposes. Thanks to their architecture, CMOS sensors are not affected by either of these two issues. And last but not least: CMOS sensors are considerably cheaper than CCD sensors. 

ITS applications

Many vision-based traffic applications work with either industrial or with IP cameras. For certain tasks both camera technologies are combined within a type of vision box to increase the effectiveness in acquiring image and video footage. Industrial (machine vision) cameras are often used for ANPR, enforcement and tolling tasks, as well as for in-vehicle applications as part of an embedded system. They provide high resolutions, at up to 2590 x 2048 pixels (PYTHON5000 from ON Semiconductor) or even 4096 x 3072 pixels (CMV12000 from CMOSIS) which make them an ideal choice for capturing several lanes instead of just one lane in a tolling application, for example. They allow for optimal precision in capturing vehicles travelling at high speeds, as is essential for free-flow tolling or speed enforcement. Their USB3.0 interface ensures secure and stable data transfer with a high bandwidth of up to 350MB/s and very low CPU load. This is of particular importance for embedded systems which typically have to make do with limited CPU resources.
High sensitivity paired with high quantum efficiency and broad dynamic range is essential in low-light conditions when the license plate and the driver of a vehicle must be captured. These properties make sure that highly reflective areas in an image (such as the license plate) are depicted with as much detail as the dark area around the driver inside the vehicle. 

IP cameras are often used in regular traffic surveillance for manual reviewing. The video stream quality is very limited as a result of the small rolling shutter sensors. Therefore more sophisticated applications which require crisp images for vehicle identification cannot be addressed with this kind of camera.

Equipping IP cameras with the latest generation of CMOS sensor technology, such as with Sony’s IMX174, opens new dimensions in terms of image quality in enforcement and traffic monitoring - 7094 Basler’s latest IP camera (the BIP2 1920-30c) is fitted with this sensor. Besides Full HD resolution and global shutter technology, it offers a real-time trigger for compressed and uncompressed individual images, which can be created parallel to a video stream thus enabling a single camera to perform tasks such as ANPR alongside traffic monitoring. The combination of these typical IP camera features and a very sensitive high-end sensor make these cameras extremely versatile tools for a wide range of traffic applications.

Again, the sensor’s broad dynamic range ensures that a single exposure of a license plate is sufficient in ANPR applications whereas traditional IP cameras’ dynamic range often required multiple exposures – sometimes resulting in license plates becoming unreadable in moving traffic. This clearly illustrates the modern CMOS sensor’s advantages in ITS applications.

  • About the Author: Enzio Schneider is market manager ITS, at Basler AG and provides an experienced resource for information and advice related to intelligent traffic systems.

Related Images

For more information on companies in this article