Advanced Imaging


Advanced Imaging Magazine

Updated: July 8th, 2008 05:26 PM CDT

Matching Image Sensor Specifications To Applications


By Yves Kessener

The basic requirement of almost all vision systems is the ability to inspect “objects” both efficiently and effectively for relevant product features and possible anomalies, quickly and with high certainty.This article describes how the specification requirements of various vision components (sensor, lens and illumination) are interlinked, and determined by the “actors on stage:” the nature of the object that is observed and the specific type of effect and details that need to be monitored.

The Object

Depending on your particular imaging application, the object you are looking for could be the earth’s surface, contrast fluid pumping through a heart’s coronary artery, a misplaced lid on a can of beans, or defects in the minute patterning structures on a semiconductor surface … or none of the above. Vision systems are as diverse as the applications they serve.

The size of your object will define the field of view in your application, and subsequently put demands on the choice of imaging optics (the lens) and illumination. Whether the object is flat or more voluminous (3-D) will define the vision system’s depth of focus requirement, and put further demands on the lens (aperture) and object illumination.

The size of the structures (and defects) that the vision system needs to be able to resolve, in relation to the overall size of the object, is the determining factor for the resolution of the image sensor. Resolving a defect with a linear size of 0.1mm on an object 10cm long will need a minimal (linear) resolution of 1000 pixels. Selecting a higher resolution is not required, as this will only increase costs and put increasing demands on processing power, signal bandwidth and the resolving power (MTF) of the entire imaging chain.

Your requirements for detecting static structures vs. dynamic processes will govern your choice in selecting a system supporting either video speeds (10-30 fps) or still picture operating modes. Specifically, in machine vision, where the number of inspections per second is key in optimizing product throughput, higher camera/sensor frame rates are very important in reducing the overall “cost-of-quality-inspection.” For the observation of ultra high-speed phenomena, e.g. in car crash tests or ballistic experiments, specifically designed systems are capable of delivering frame rates in millions of frames per second.

1 2 3 4 next

Subscribe to our RSS Feeds