Cameras installed everywhere aren't there simply for watching you. They exist to help people, systems, and machines make decisions and take action. For machine vision, more megapixels aren't necessarily the answer.
Intelligent traffic systems (ITS) deploy a number of cameras. Some are there to monitor and control speed, traffic flow, and toll collection remotely. Others automatically recognize license plates.
Because they're installed outdoors, these systems need to see in through rain, in the dark, or in bright daylight. Reflections of images on a rainy street can confuse cameras. That's when high optical dynamic range is required for image sensors.
Cameras used in ITS, by nature, need to capture images of moving cars, making a CMOS sensor with a global shutter absolutely critical.
Rolling shutters (used predominantly on CMOS sensors) capture an image by scanning across the frame gradually. Though the whole frame is displayed at the same time during playback, not all parts of the image are recorded at exactly the same time. The effect of a rolling shutter on an image might be negligible, but not if the camera or object moves during the exposure. That's why regular CMOS image sensors with a rolling shutter don't work for ITS applications.
A global shutter -- in which the entire frame is exposed in the same time window -- prevents uncorrectable motion blur. In a CMOS image sensor with a global shutter used in ITS applications, the expected pixel noise levels are below 10 electrons. The expected shutter efficiency is 99.9% or better.
Next Page: Rice-sorting machines