How does vision inspection work?

Just like a superhero, automated vision inspection systems are faster than a speeding bullet — or at least, faster and far more reliable than the human eye. So how do they do what they do?

There are four basic stages in machine vision. The first two are about the images, and the second two around sorting: 

  1. Image acquisition 
  2. Image filtering 
  3. Feature extraction
  4. Image processing 

Let’s go through them… 

  • Acquiring the image: This is where it all starts, so it’s vital the camera is set up the right way — position, lighting, etc. — so all features of interest (label the right way, code present, pin holes, surface texture and so on) can be captured properly. After the camera takes an image, it’s converted to a digital format and put into the system’s memory. 
  • Image filtering: To enhance the elements of the image that are important to the inspection process — i.e., what you’re inspecting for — the processor uses various algorithms to segregate the desired features from all the rest of the image. At the same time the system effectively ignores (or filters out) aspects that are not relevant for the inspection i.e. background noise 
  • Extracting features: This is all about quantifying the relevant image features in terms of measurements – the distance between two points, the size of a hole, the number of holes etc. So for this step, the processor firstly identifies, then quantifies, critical features in the image (e.g.: label position, a code is present, the surface finish and so on).  
  • Image processing: To “recognise” or “classify” features, the system allocates qualities with the same properties to a predetermined set. This could be for flaws, say with labels upside down or presence of a Use By code; or for the correct appearance, say the right barcode or tamper seal intact. 

Machine vision does all of this faster than a speeding bullet and more powerfully than a locomotive. They can’t quite leap tall buildings in a single bound, but vision inspection systems are a bit like Superman in that they’re superhuman. 

Want to find out more about us? 

1 stars2 stars3 stars4 stars5 stars

1 stars2 stars3 stars4 stars5 stars