HDR Cameras and Their Applications in Embedded Vision
Embedded cameras come with a wide variety of features such as high resolution & frame rate, low light sensitivity, NIR performance, global shutter, etc.
Of these, High Dynamic Range (HDR) is critical in ensuring the functioning of camera modules in a wider spectrum of lighting.
This is because many cameras have to operate in diverse settings – say both day and night. Having a wider dynamic range helps in such scenarios.Today, we discuss in detail what dynamic range is, how HDR cameras operate, and applications where they are commonly used.
What is Dynamic Range?
In simplest terms, dynamic range is the difference in maximum and minimum light intensities of a scene or image, taking into account noise. It can also be defined as the difference between the luminosity of the darkest and brightest, or saturated, areas of a scene. Dynamic range in cameras is usually represented in dB. The higher the dynamic range of the camera, the better its ability to capture details in brightly lit and dim areas of an image.
What is High Dynamic Range and Why is It Important?
Normal Camera vs. HDR Camera Output
The issue is that the normal linear dynamic range of an image sensor is not enough. For example, many scenes have dynamic ranges that exceed 100dB and the human eye can perceive 90 dB.
Most consumer CMOS image sensors might have a linear dynamic range of around 48-50 dB. This is called linear because the sensor response to light is linear over a specific exposure (or integration) time.
Higher-quality image sensors, such as the onsemi AR0521, have a better linear dynamic range of around 74dB. This is due to the better noise performance of the sensor. High dynamic range sensors, such as the onsemi AR0821, have a dynamic range as high as 140 dB.
Now, why are HDR cameras needed in embedded vision applications?
Imagine this. In a smart traffic system, you have a camera with a bright sun in its field of view. As vehicles pass through the field of view, the camera has to capture their images with the sun in the background. In such a scenario, there is a high likelihood of a portion of the image getting washed out if you use a normal camera. An HDR camera can overcome this challenge by creating an even intensity of light across the image despite a certain section of the scene being overly bright.
One thing to note here is that, though high dynamic range cameras can effectively capture images under limited light, for extreme low light conditions, it is recommended to go with a specialized low light camera module. This is mostly because the sensor itself has to come with a different architecture to make imaging in low light possible. In some cases, the sensor also needs to have NIR sensitivity to be able to capture images in night conditions when IR illumination is used.
Working Principle of HDR Cameras
For a camera to have HDR capabilities, the sensor and ISP (Image Signal Processor) have to support HDR. But there is more. HDR cameras work differently compared to ordinary camera modules. Let’s explore that in this section.
This is the most popular method of obtaining HDR images. When we say multi-exposure, in most cameras, several exposures are used – one for low light, one (or several) for mid-tones, and one for bright light. These images are then combined to obtain a single image with all the necessary details. This technique is also referred to as DOL HDR (Digital Overlap HDR) or image stacking.
In some cases, an even higher number of exposures are used, such as nine-exposure HDR, in some applications where high levels of details are required in dim or very bright areas of an image. This can capture the entire range of luminosity in a scene.
The single-exposure HDR method relies on data from a single image to mimic the effects of the multi-exposure HDR method. The raw images from a camera can be enhanced using software to obtain details that might not have been otherwise visible. This method is effective when the dynamic range of the image is not very high.
There is also a technique used to increase the dynamic range of a sensor by changing its well capacity during integration time, but this is not commonly used.
How to Choose the Right HDR Camera for an Application?
Just like how you would select a normal camera, the process of selecting an HDR camera for an embedded vision application is similar, only that you would consider the quality of the HDR output as an additional factor.
While evaluating a camera for HDR performance, a best practice is to check its output by integrating it into your end product or prototype rather than blindly going by the HDR value. This is because the quality of the HDR output not only depends on the sensor but the ISP and interface used.
Additionally, as in most cases, the effect on dynamic range performance depends on the number of exposures, this can result in a decreased frame rate. For example, an onsemi AR0821 sensor, with built-in HDR processing, can support full frame rate HDR processing with 3 exposures at up to 40 fps, while 4 exposures reduce this to 30 fps. If a fewer number of pixels are processed (for example if the resolution is reduced through binning/skipping), then the frame rate can be increased while maintaining or improving HDR performance.
Once the HDR quality is assessed, depending on the application, you need to look at features such as the following (as in the case of a normal camera):
- Resolution and frame rate
- Shutter type
- Chroma type
- Lens mount
- Interface and connector
- Field of view
- Focus type
- Low light and NIR sensitivity (if relevant for the application)
There might be a few other niche features that need to be considered – such as enclosure, depth range, sensor size, etc. – depending on the end application. But the above are the most important ones you need to factor in.
Applications of HDR Cameras
As discussed above, HDR cameras are used when the range of lighting of the scene is high. The most common use case is when the scene often contains very bright areas (saturated) that cannot be captured using a normal camera along with important low-intensity areas. Let’s dive into the details of some key applications where HDR cameras are recommended.
Smart Traffic Systems
We touched upon this briefly earlier. Intelligent traffic systems are used for vehicle counting, rule violation detection, people counting, and even facial recognition of passengers.
These camera-based devices are often placed in traffic junctions or tolls to get a complete view of the scene. Given that they have to capture images in bright lighting conditions, HDR cameras are recommended.
Outdoor Robots & Autonomous Vehicles
Robots & vehicles used in outdoor environments such as delivery robots, agricultural robots, drones, and autonomous tractors & lawnmowers often have to operate in bright lighting. This necessitates the need for having a camera that prevents the loss of details from the output image. HDR cameras meet such a need by retaining the details of the bright as well as dim areas of a scene.
Parking Lot Management Devices
Parking lot management systems use cameras for similar purposes as smart traffic devices.
They help identify vehicles through automatic number plate recognition. They are also used to find empty slots in a parking area to guide incoming vehicles to easily find a space to park.
Cameras are used in open parking lots as well as closed ones.In the former, there is no doubt that HDR cameras are required. In the latter, if the camera is placed at the entrance and it has to face bright light at certain times of the day, or to avoid being ‘blinded’ by vehicle headlights, an HDR camera might be required.
Surveillance systems used in streets, retail stores, government buildings, etc., are often placed in locations where certain areas of their field of view are masked by bright or dim light. HDR cameras are a lifesaver in such scenarios.
Biometric and access control systems used in airports, borders, and office buildings have to operate in extreme lighting conditions at times.
Given that accuracy is of paramount importance in these systems, a normal camera might not suffice in case there is a high variance in the luminosity levels of the environment. HDR cameras are a perfect choice for an application like this.
Though we looked at a set of applications where HDR cameras are recommended, it is difficult to generalize and say that they are needed in all scenarios where such systems are used. For instance, if a surveillance system is placed in such a way that it never has to face bright or dim light, an HDR camera might not be required at all. So, be sure to make your choice based on the uniqueness of your application. It is a good idea to consult a TechNexion camera expert for this.
HDR Cameras from TechNexion
TechNexion aims to help its customers build modern embedded vision systems by offering solutions such as camera modules, system on modules, panel computing systems, and more.
HDR cameras are one of the most important weapons in our camera arsenal. Each of the camera solutions comes with different variants with changes in lens mount type, interface, and connector.
Please visit the embedded vision cameras page to learn more about TechNexion’s camera solutions. Another way to get a quick overview of our camera solutions is to watch the video:
Get a Quote
Fill out the details below and one of our representatives will contact you shortly.