Measuring parts is a very common type of inspection. It can be done without contact, and at high speeds. We frequently measure in the range of ten millionths of an inch but the accuracy and precision of such activities is virtually unlimited. Successful machine vision results are essentially a function of matching the hardware, software, and optics with the constraints generated by the machine vision inspection.

High Precision Measurements in Machine Vision Today the accuracy and precision on vision systems seems virtually unlimited. This particular example provided the following results with little effort.

• Measurement Precision and Accuracy to .000010”
• 30 Measurements and 12 calculated outputs in a cycle time of less than 67 milliseconds
• 45 Separate Programs for Different Parts

Today it’s possible to measure some parts in as little as 4 milliseconds.

Distinguishing between good parts and bad can be done with a multitude of approaches, from simple pattern searches, blob or stain detection, differentiating by color, custom vision tool application, measurements, and custom programming.

This customer had an existing system that needed programming improvements to achieve their desired system performance. We helped them to add specific vision tools to achieve an optimized result. Do you have an existing system that needs a few programming additions to move to a new level of performance?

Occasionally parts will escape a screw machine operation where a threading tool has broken. This system finds any parts that move downstream with missing threads and ejects them from the process.

The picture on the right shows an aluminum burr in the bore of a machined part. The burr is only .002” thick. We can assist in finding defects on parts which are sometimes difficult to detect with the human eye.

There is an old saying in quality circles that “100% inspection is only 80% effective. The basis of that adage is quite simple. No matter how easy and obvious the task a person will eventually lose their focus and a bad part will get through. In fact the simpler the task the higher the potential as the perception of simplicity leads to complacency.This operation looks for a copper washer assembled on a part and rejects all parts without the washer.

Line lasers can complement vision systems to allow for a variety of inspections. In addition to the obvious uses in measuring vertical height differentials they can also be used to look for surface non-conformities and dents in a variety of materials.

Pattern matching was one of the first tools developed for vision systems. In its most basic form it is simply comparing and counting pixels against a master file.  Pattern matching by itself has probably been responsible for selling more vision systems than any other software feature. It is incredibly simple to set up and demonstrate to a customer. When looking at a couple of sample parts in a quick demonstration it can promote a deceptive confidence in this vision tools ability to solve the problem at hand.Then when a customer integrates a camera in their process and they start looking at hundreds, or thousands of parts the weakness of using this tool exclusively becomes very apparent. There are very specific constraints which must be applied to use such a tool successfully.Clearly pattern matching is an important tool when applied correctly. Used in combination with other vision tools it can be utilized to develop very strong solutions in vision system integration.