|
|
|
The simplest version of a Machine Vision System (MVS) consists of a light source, camera and computer software that performs fast analysis on digitized images. It can be looking for location, size flaws, etc. Unlike the other types of image analysis, the MVS also includes a mechanism that immediately reacts to the images which do not meet the parameters stored in the computer. For example, defective wafers are flagged and sent back for reprocessing. MVS functions include location, inspection, gauging, identification, recognition, counting and motion tracking. These systems do not necessarily need to operate at any given standard frame rate. The MVS can process limited data much faster than any human. Many operations can be performed faster, cheaper and more accurately by machines. MVS can operate 24 hours a day without fatigue. MVS offers consistency that does not come naturally to human operators. And finally, MVS can operate in harsh environments that many human inspectors would find unacceptable working conditions. Because the MVS is making unguided decisions at very high speeds, it must receive an almost perfect image data to work with. Because of the high speeds of the operations, only limited image correction can be applied. This highlights the need for very high linearity over the entire dynamic range of the sensor, as well as very small pixel to pixel cross talk for the sharpest possible image. Adept's CMOS image sensors are an ideal choice for the MVS. Our proprietary designs deliver linearity >99.5%, pixel to pixel cross-talk <2% up to 800 nM, PRNU's under 1% and FPN under 0.2%. Temporal noise is below 15e- rms for custom pixels. Patent pending low power on-chip ADC architectures provide resolutions up to 14 Bits at video rates for sensor formats up to 1.3 Megapixels. Adept CMOS is the clear choice.
|
|
Copyright © 2002 Adept IC Design
|