Surveillance and situational awareness systems are relying on daylight full motion video more and more. The camera requirements for this use are unique and cannot be met by consumer cameras or machine vision cameras.
Applications such as border surveillance or situational awareness systems involve a human viewing the images so the camera performance needs to resemble the human eye in real-time as much as possible. The human eye is not linear – it is adaptive to light conditions and has high dynamic range. The way this is recreated in the camera needs to be examined though, as you do want to retain detection and recognition capabilities from long distances.
Rugged full motion video cameras are available and allow you to have high sensitivity and high resolution at 30+ fps. They include optimized functionality to adjust for changing lighting conditions and other atmospheric conditions such as fog and other image processing needed for viewing such as accurate color reproduction and automatic white balance. There are true rugged options that can provide 24/7 operation on airborne platforms. The cameras can be combined with zoom optics for even better identification performance.
In contrast, many consumer cameras can reproduce the scene for viewing very well, even cell phone cameras, with limited sensitivity. They often use special tricks such as taking 2 pictures with different integration times and merging them together to get better dynamic range. But, if the camera is moved a little bit between the 2 shots – there is a loss in sharpness and motion blur. There usually are good algorithms in the cameras to merge the images together while eliminating the blur, but this is not possible at 30 fps. These are excellent for personal use, but not designed well when you need real-time frame rates and the images provide critical input information such as threat identification in global security applications.
One of the other issues with consumer and security cameras is that the lens options are limited. The sharpness in the image is a combination of the number of pixels over the target and the optics. Without a well-matched lens (measured in Modulation Transfer Function or MTF), you lose the ability for identification from the images.
Similarly, machine vision cameras are also not ideal for these applications as they are optimized for the pixel information to be used for measurement inside an instrument. All of pixels must behave the same behavior for accurate measurements so the most important camera parameters are linearity and uniformity. This must be combined with high full well capacity and fast frame rates for optimal performance. No image processing capabilities are included to deal with changing lighting conditions or extreme environmental changes.
As an additional note, when selecting a camera, consider whether the camera manufacturer creates their own image processing or uses a standard chip set. There is typically a cell phone manufacturer behind a standard chip set. The cell phone manufacturer provides the volume, and if they change their mind, production is stopped and this functionality is no longer available to the industrial camera market. If the camera provider has their own knowledge/IP, the technology can be supported as they control the supply (and can adapt it to specific system challenges).
Click here for information on Adimec’s rugged HD full motion video cameras with our own image processing and contrast enhancement.