Techno requirements for machine vision
Most notable is that the new Python family -- designed to offer hardware scalability (pin compatible) and high configurability -- keeps up with all the emerging technical features required for new industrial CMOS imaging applications.
Yole analyst Mounier lists high dynamic range, near-IR capability, low light sensing, and global shuttering as the key “techno requirements” for machine vision. High dynamic range means that “cameras must extract information out of a non-saturated image and must compensate for rapid illumination changes (in ITS, or reflective surface inspection).”
The trend toward night-vision in high-end surveillance and ITS will demand “near-IR capability,” he notes. IR capability is also needed “in some specific applications like solar panel inspection.”
By “low light sensing,” Mounier means that a “sensor should operate in a light-starved environment for high-speed imaging, and for making lighting economies.”
Last but not least, CMOS image sensors require “global shuttering.” Fast moving objects should be imaged without motion distortion. “Global shutter is a must-have feature for all high-speed applications!”
System block diagram of Python family of image sensor solutions
All building blocks in green can be implemented with ON parts.
(Source: ON Semiconductor)
ON Semiconductor’s Ringoot says CMOS image sensors in the Python family combine low noise and high sensitivity with high-speed global shutter imaging.
Company patents leveraged in Python include in-pixel CDS technology enabling global shutter imaging with Correlated Double Sampling in a compact pixel size. The result is optical performance “close to on-par with that of CCD,” says Ringoot.
Further, other new pixel technologies introduced to the Python family include: improved noise performance, dynamic range, and parasitic light sensitivity (PLS) and responsivity. Under the PLS structure, the original image value is kept in the pixel. Because the value is not stored in a perfect binary memory, but rather in the voltage domain, Ringoot says, “There may be leakage. Our new technology, however, is designed to minimize that leakage.”
The new Python pixel, according to the company, “allows the capture of fast moving scenes without distortion by combining a read noise of less than 9 e-, with 7.7 V/lux sensitivity and frame rates as high as 850 frames per second.”
CMOS image sensors in the Python family offer operational support across the -40°C to +85°C industrial temperature range. (Note: The temperature range required for the automotive sector, however, is even wider, ranging from -40°C to +105°C, according to Ringoot.)
Among a number of industrial applications the Python family of CMOS image sensors are designed for, ITS might be a good example to illustrate key technical features required in the new generation of image sensors. Ringoot explained that ITS needs “global shutter, 60 frames per second, and be capable of capturing a number of pixels per line that ranges from 1,280 to 2,000.”
It must work in “less optimal light conditions, and in the industrial temperature range (in extreme weather).” After all, one ITS camera is expected to watch at least two tracks on a highway, working in all weather conditions all day, and it needs to be able to capture license plate numbers.
ON Semiconductor takes pride in the ease of integration of CMOS image sensors offered in the Python family. Aside from a variety of pin- and configuration-compatible solutions, ON Semiconductor provides FPGA-based reference kits and schematics, so that designers can test which CMOS image sensor is best for any specific application, says Ringoot.
— Junko Yoshida, Chief International Correspondent, EE Times