Avigilon on adaptive IR

Lighting is one of the most important factors in image quality. As it gets darker, a surveillance camera's ability to capture evidentiary detail such as faces and licence plates rapidly decreases. In order to overcome this, artificial sources of light can be added to the scene, the most common of which is IR illumination.

While the human eye can detect visible light in the ranges of 390 to 750 nanometer (nm), IR light is at a higher wavelength (typically 850 nm), making it invisible to the human eye while providing the camera the ability to “see.” Since IR light has a higher wavelength than the color spectrum, the resulting image will be in monochrome (black and white), but can capture crucial detail in complete darkness.

However, using IR as a source of illumination has a few side effects. For example, the IR illumination must be calibrated to the camera's field of view to provide the most effective illumination over the entire imaging area and remove unwanted side effects such as hotspotting and dark edges. This calibration is especially problematic when using a camera with a varifocal lens, as the camera's field of view changes based on the position of the lens' focal length. Another challenge of using IR as a source of illumination is that objects that come close to the source quickly become oversaturated and evidentiary detail can be obscured.

Advances in camera technology, notably adaptive IR technology, can help overcome these challenges and improve the quality of the evidence captured in the image. This paper provides an overview of the adaptive IR technologies currently available, their purpose, and their short comings.

Adaptive IR technology
There are many cameras in the market, both analog and IP, that offer integrated IR.However, almost all of these cameras provide only static IR settings that cannot be changed to meet different environments.

Avigilon's Adaptive IR technology combines multiple adaptive processes:
Content Adaptive – Optimizes camera settings dynamically based on changes in the imaging area to improve the quality of the evidence captured through two methods:

Oversaturation Detection: Removes oversaturation while reducing motion blur and providing more effect license plate illumination.

Ignoring Background Oversaturation: Ignores stationary oversaturated areas to provide the most detail of moving objects, even in the darker regions of the image.

Zoom Adaptive
When a camera comes with a varifocal lens, for example a three-nine mm lens, the user has the ability to change the field of view of the camera, narrowing the size of the area covered as the camera is zoomed in.

When using IR light to illuminate the imaging area, a camera with zoom adaptive IR technology will dynamically focus the IR light to the area covered by the camera. As a result, as we zoom in, the IR beam angle is focused along with the lens, to illuminate an area that is smaller but farther away. Without zoom adaptive technology, the manufacturer sets the camera's IR illumination to a pre-set beam angle, covering a fixed distance and area, generally somewhere between the two ends of the lens' focal length.

When IR settings are fixed, the illumination area cannot be changed based on the camera's field of view. As a result, static IR only provides optimum illumination when the camera is at the same zoom position as the IR illumination area. If the camera's field of view is wider than the illumination area (the camera is zoomed out further), edges of the imaging area will not be illuminated and important detail can be lost. On the other hand, if the camera's field of view is narrower than the illumination area (the camera is zoomed in further), IR light is being wasted on areas not captured by the camera. As a result, less light is available for the imaging area. This creates darker, noisier, and less-detailed images.

Zoom adaptive IR ensures uniform illumination over the entire imaging area at every zoom position, and eliminates wasted IR by focusing all the available IR to the camera's imaging area.

Content Adaptive
While zoom adaptive technology provides the most effective illumination based on the angle of view of the camera, during operations, the camera's imaging area can change dramatically. Changes can come from either subjects moving through the scene or from other sources of illumination. Avigilon's content adaptive technology dynamically adjusts the camera's settings based on the changes within the imaging area to improve the quality of the evidence captured through the two methods of oversaturation detection and ignoring background oversaturation.

Oversaturation Detection
At any given time, a camera can “see” objects within a specific range of lighting conditions. Let's say you have a light source in the corner of the room and three objects at different distances from the light source.

Automatic Settings
The camera's automatic mode will set the camera's settings based on the average scene illumination. The most likely result is that the second object is clearly displayed while the other two objects are either oversaturated or completely dark.

Shortest Exposure
Some camera features, such as backlight compensation and exposure, allow the user to bias that range towards darker or brighter lighting conditions. For example, the camera's shortest exposure will likely display the first object clearly but leave the other two objects completely dark.

Longest Exposure
The camera's longest exposure will likely display the last object clearly but leave the other two objects oversaturated.

WDR
Cameras with WDR can see objects within a larger range of lighting conditions. However, this range is still limited as it clearly displays the first two objects, while leaving the last object in the dark.

Content adaptive technology adjusts the image brightness to the part of the image that is of most interest (moving objects) by ignoring the parts that are least interesting (stationary oversaturated objects such as street lamps).

Since WDR cameras can see a wider range of illumination, a larger portion of objects within the scene will have appropriate brightness and therefore adjusting the image brightness for specific objects will have a smaller effect on the overall picture. On the other hand, cameras without WDR have a much smaller range of illumination and adapting the image brightness for specific objects is likely to result in a much bigger difference in the image.

While cameras with WDR feature will still experience the benefits of Content Adaptive technology, those cameras without WDR feature will experience a larger benefit in similar lighting conditions.

Achieving the right level of illumination on objects of interest is particularly important when using IR illumination at night or in dark environments. Since the general environment is extremely dark and the source of illumination is extremely bright, oversaturation can happen quickly when objects come closer to the camera and the camera is trying to balance what it can see.

Oversaturation is further heightened by the fact that different materials have different light reflectivity properties. For example, license plates and safety vests have highly reflective coatings that will oversaturate the object and its surroundings. Cameras without content adaptive technology will set the camera's imaging settings based on the average illumination in the imaging area. Since most of the imaging area is extremely dark, the average will allow important objects to be oversaturated and detail to be missed.

Cameras offering content adaptive IR technology analyze the relative size of the oversaturated area and will adjust the camera to remove the oversaturation when the area occupied by the object is large enough. This technology effectively prevents objects closer to the camera from becoming oversaturated (since the object occupies a larger area in the image) and ignores oversaturation that is caused by a light source (car lights or street lamps in the background). It is a very effective way of ensuring that the objects you care most about have the right level of illumination and detail.

Content Adaptive Technology (Avigilon HD Bullet Camera)
There are two predominant ways to remove oversaturation, each with its own strengths and side effects. The first way is to decrease the brightness of the IR illumination. Since the object is most likely oversaturated as a result of the camera's IR illumination, decreasing the IR illumination is an effective way of removing the oversaturation.

The second way is to reduce the camera's exposure. Exposure controls the amount of time that the sensor is exposed to light. Reducing the exposure reduces the amount of light entering the sensor for any given frame, and as a result effectively removes the oversaturation. The important thing to note is that in both of these cases, the overall image will be darker, as in the first case the amount of light illuminating the scene is reduced, and in the second case, the amount of light entering the sensor is reduced.

However, since you generally care more about objects closer to the camera rather than background objects, this is a very effective trade-off in the surveillance industry. Each of these methods of removing oversaturation, namely decreasing IR illumination and reducing the camera's exposure, has its own advantages. If there are other sources of light brightening the background of your scene, reducing the IR illumination allows the image's background to remain as bright as afforded by your other light sources. On the other hand, when you reduce exposure, the entire image will be darker.

There are also a number of advantages to reducing exposure rather than IR illumination. First, high exposure results in blurriness when objects in the scene are moving; therefore, reducing exposure provides sharper images of moving objects within the scene. In surveillance, the objects you generally care most about are those that are moving; for example a person entering the scene or a car going through it. Lowering exposure can make a large difference in both of these cases as motion blur can hide facial detail and obscure license plates.

Another advantage of reducing exposure rather than IR illumination is evident when capturing a license plate on a car, even if the car is not moving. As shown in the diagram below, a car's license plate is located between two headlights that are very bright sources of illumination. However, since that illumination is shining away from the license plate, the plate itself is not being illuminated. (Figure images courtesy of Avigilon.)