Finding the right camera for an ITS can be daunting, but integrators and OEMs can find success by following these guidelines.
Intelligent transportation systems (ITSs) improve transportation safety and mobility, enhance productivity, enforce laws, and ultimately help generate revenue. However, failure to consider harsh environments and specific application needs when choosing cameras for an ITS will impede system success. Additionally, an overwhelming number of options make it difficult to know where to start.
Cameras are prominently used in ITS technology in applications such as traffic monitoring, automatic license plate recognition (ALPR; referred to as automatic number-plate recognition, or ANPR, in the UK), access control, high-occupancy-vehicle (HOV) lane monitoring, parking enforcement, speed enforcement, and asset protection. Depending on the application, different types of cameras and systems may be required. All deployments pose challenges, whether the camera setups require long cable runs and the ability to withstand tough environments or need embedded systems with edge processing capabilities for quick feedback. Careful consideration of these factors can ensure system success, lower cost of ownership, and future proof operations. This article presents common challenges and methods to overcome them and help position ITSs for success.
Start With Image Quality
When researching cameras for ITS applications, all planning must begin with image quality. In the case of ANPR/ALPR applications, the system has two main tasks: recognizing that a number /license plate is in the camera’s field of view and then decoding the plate. Both tasks are subject to error. Successful ANPR/ALPR systems require high-quality images for the software to perform optical character recognition (OCR) and optical character verification (OCV) tasks. System designers and integrators need powerful cameras.
Cameras with low-quality image sensors will not produce high-quality images, so the image sensor must be a key consideration. ITS scenes may vary greatly in brightness due to changing outdoor lighting conditions and a camera must be able to capture and provide data and details to the brightest and darkest parts of an image and variations in between regardless of lighting conditions. The measurement of a camera’s ability to detect higher and maximum and minimum of light intensities is called “dynamic range”; the higher the decibels (dB), the better. High-quality image sensors such as Sony’s Pregius and Pregius S line of global shutter CMOS image sensors deliver high dynamic range sensitivity, making them ideal for a variety of ITS applications.
Teledyne FLIR’s Blackfly S camera family offers an impressive range of camera models based on Sony Pregius and Pregius S image sensors. These range from 1.6 MPixel to 26 MPixel, making them suitable for a wide array of different ITS tasks. The BFS-PGE-161S7C-C (color) and BFS-PGE-161S7M-C (monochrome) cameras, for example, feature the 16.1 MPixel Sony IMX542 Pregius S sensor offering 70.46 dB dynamic range, 9609 e- saturation capacity, and quantum efficiency measurements of 45.76% at 470 nm (blue), 52.26% at 525 nm (green), and 33.49% at 630 nm (red). Purchasing a high-resolution camera when the application does not require it adds cost onto the total project but makes sense if considering future scenarios. Integrators and OEMs, for instance, can future proof operations to allow for road expansion and avoid having to change out lower-resolution cameras later.
Capturing Color, Overcoming Glare
Color is another important quality metric. In the case of ANPR/ALPR, while black-and-white images show the number/license plate, color images provide context on what is occurring in a given situation (traffic lights, color-coded road signs, and so forth). Color is also important in situations where cameras mounted on vehicles check for improperly parked cars. In Spain, for example, blue lines on the street indicate that a driver must pay for the parking spot, while green lines indicate that the spot is for someone who lives in the neighborhood. Embedded vision systems can automatically check for infractions but need high-quality color reproduction to do so.
Image sensors have a particular response to lighting (quantum efficiency), and each lighting condition, such as sunlight, has its own emission spectrum, impacting how an image will appear when captured. Quantum efficiency determines an image sensor’s ability to convert photons into electrons and varies depending on the wavelength. Color collection tools factor in how each color channel interacts with the others and independently scale each color channel. A color corrective matrix measures and compensates for these interactions to more accurately reproduce the real-world colors of a given subject. This is particularly important in applications where minor differences in color can negatively impact the accuracy and reliability of the results.
Cameras and vision systems may at times struggle when dealing with reflections and glare on reflective surfaces such as glass. For example, if an ITSs wants to look into a car for HOV lane monitoring, reflections and glare from the sun may prevent the camera from capturing images of the car’s interior. Teledyne FLIR offers two cameras (one GigE, one USB) based on the 5 MPixel Sony IMX250MZR monochrome polarization sensor and one camera (USB3) based on the 5 MPixel Sony IMX250MYR RGB polarization sensor.
Figure 2: Polarization cameras can capture images of the car’s interior even in difficult scenarios where glare or reflections may be present.
Find a Hardware Expert
Some integrator and OEMs in the ITS industry may have more expertise in software as opposed to hardware. Choosing, testing, and optimizing hardware presents a challenge, but an experienced camera vendor can provide technical consultation on the project and aid with selecting and setting up the cameras. The vendor should also provide referrals to trusted partners when dealing with accessories and services not offered by the company, including lenses, cabling, housing, and software development.
OEMs and integrators also need support through the design and development phase, as well as help with system setup and software advice. Users should seek out camera companies that provide these services through a combination of system engineers and support teams with frequent check-in calls. Prior to choosing a hardware provider, end users should also ask about system support as well as how the provider handles customer relations.
Striking a Balance: SWaP-C and Performance
A popular term in research and development as well as military applications, SWaP-C stands for “size, weight, power, and cost.” Applications of all types require devices, systems, and programs with optimal SWaP-C. ITS applications are no different. Board-level, low-cost ITS cameras have grown in popularity in recent years. Cameras coming in smaller packages aren’t enough, however. These cameras must strike a balance between SWaP-C and performance. One way Teledyne FLIR balances this is by guaranteeing that all its board-level cameras are identical in feature set compared to the cased versions of the camera.
Figure 3: Despite the compact design, Teledyne FLIR board-level cameras offer the same feature sets as their cased counterparts.
One ITS trend involves deploying higher-resolution cameras to cover multiple lanes on a highway. Integrators that previously used board-level cameras with low-resolution image sensors may look to upgrade a 1.3 MPixel camera with an 8.9 MPixel or 12 MPixel camera, for example. If the new camera does not have the same form factor as the previous model, however, the integrator will need to redesign and recertify it. Teledyne FLIR offers board-level cameras with the same form factor — in a variety of resolutions — so integrators can much more easily upgrade systems.
A Shift Toward Embedded
As with the machine vision market, ITS has seen a shift toward embedded systems with edge computing capabilities in a low-power, small form factor design. The application that may first come to mind when one considers embedded systems in ITS is in-vehicle deployment, but embedded systems can suit nearly any ITS or smart city application.
Integrators should evaluate some of the most popular embedded hardware options for a potential fit. For example, can the cameras work with NVIDIA’s Jetson TX2 or Xavier embedded modules? Many ITS applications involve sophisticated algorithms and require a system capable of processing them.
To that end, Teledyne FLIR launched the Quartet™ Carrier Board for TX2. Specifically customized for ITS applications, it allows customers to directly connect four board-level USB3 cameras to the TX2 without the need for hubs or converters. Each connector in the power-over-cable board has its own bus, so it does not need to share bandwidth with other connections. As an example, with the Quartet, integrators can simultaneously deploy a high-resolution color camera for overall context, a monochrome camera for ANPR/ALPR, and a polarized camera to see through windshields — all in a single connected system.
Figure 4: Designed for the Jetson TX2, the Quartet Carrier Board can connect to four cameras in space-constrained applications.
Rugged, Reliable Design
Cameras deployed into ITS applications must be able to handle the task from a physical standpoint. In the case of on-vehicle applications, integrators must consider the camera’s ability to operate in extreme temperatures, for example. While many cameras are integrated into a protective housing to withstand the weather, cameras still need to work in in hot temperatures (exceeding 50 degrees Celsius); Teledyne FLIR makes sure all its camera models have successfully gone through HALT tests (Highly Accelerated Life Testing), ensuring that no camera failures occur from -30 to 80 degrees Celsius.
Integrators also need to consider shock and vibration when choosing a camera. Cameras must be compliant with industry specifications for shock and vibration to ensure image quality and long-term system reliability. When purchasing a camera, integrators should research what type of testing it has undergone. Vibration testing done on Teledyne FLIR cameras is documented publicly.
In general, ITS cameras must be extremely reliable. Borrowing a platitude from the sports world, “the best ability is availability.” Cameras must be able to perform the required tasks for a long time without malfunctioning or breaking and needing to be replaced. All ITS integrators know the cost and hassle involved in switching cameras in a system that is already deployed. Integrators can avoid the inconvenience and embarrassment by choosing high-quality cameras proven to perform in an ITS environment for many years.
Leveraging Time Stamps and GPS Data
GigE Vision cameras are popular in ITS applications for several reasons, including their ability to support extremely long cable lengths. Another, perhaps slightly lesser-known, reason is their ability to support the IEEE 1588 precision time protocol (PTP). Cameras supporting IEEE 1588 PTP accurately time-stamp images at the point of exposure. Additionally, the standard provides advanced capabilities, such as allowing multiple cameras to execute synchronized image acquisition based on an internal time-based command, without the need for external triggering.
Figure 5: Path delay is calculated and factored into synchronizing the clocks between devices. The primary sends two signals to the slave (1) and (2). The secondary then sends a signal back (3) and path delay is calculated and applied to sync the clocks (4).
This standard is important because it provides the ability to synchronize to external hardware and embed GPS data in image streams. An example of this is accurate detection of vehicles in violation of the speed limit (without radar). Time stamps from two different points can help determine whether a vehicle exceeded the speed limit, and accurate image times from both points will simplify high-precision speed analysis as well.
Time stamps are also important for gantry-based automatic toll enforcement. An inaccurate time stamp will not produce an image of the whole car; the time stamp must be in sync for toll enforcement. When it comes to system design, one must consider the need for time stamping prior to purchase.
Achieving System Success
Inevitably, some cameras available on the market today will eventually fail and cause headaches — not to mention lost time, money, reputation, and public trust. When evaluating cameras, remember to consider image quality, hardware flexibility, embedded capabilities, physical design and reliability, and the importance of time stamps and GPS data. By considering these factors, integrators set themselves up for overall system success.
Teledyne FLIR offers a range of rugged, compact industrial cameras that can be reliably deployed into various ITS projects and cover low-resolution (1.3 MPixel) needs all the way up to high-resolution (20 MPixel-plus) needs for multilane coverage. Reach out to us today to learn how your ITS application can be transformed by our cameras — whether off-the-shelf or customized to your specific needs.
Figure 6: Cameras such as Teledyne FLIR's Blackfly S leverage the latest CMOS image sensor technology from Sony's Pregius and Pregius S line.