Sunday, October 7, 2012

The intelligent bacon slicer

There’s nothing like the smell of bacon frying in the pan, unless you’re a vegetarian I suppose, but when you buy a pound, or a kilo, how do you know what you’re getting?

Well the butcher, or more likely the packaging machine, weighs it. But there lies a problem. The retailer wants to give you exactly a pound of bacon, and no more. But as bacon comes in slices he has to add in that extra rasher so as to make the minimum weight.

This giveaway is a problem, but machine vision has come to the rescue.

Stemmer Imaging announced recently that they’ve been working with slicing equipment maker Marel on an automated grading system. The issue here is that bacon is sliced to a thickness, but the weight of a slice can vary depending on the ratio of meat to fat. So the “smart” slicer looks at the end face of the slab of bacon and determines how much meat and how much fat. From here it can determine the optimum thickness to ensure the final pack weighs exactly the declared weight, and no more.

Clever, eh?

Sadly, there’s little information about the IBS2000 Vision Bacon Slicer on the Marel website, but I did find a write-up on, of all places, the ITS International website. They’re the people who cover intelligent transportation systems, (another big vision market,) and if you scroll to the bottom of “Machine vision - cameras for intelligent traffic management,” (October 2011,) you’ll find a sidebar piece about the bacon slicer.

But the story doesn’t end there. While Googling bacon and vision, up popped a link to a University of Nebraska-Lincoln report for the National Pork Board. Published in 2000, “Quality Lean Growth Modeling-Bacon Quality Assessment,” (it’s a pdf,) includes a section titled, “CHAPTER 3- MACHINE VISION ANALYSIS OF BACON”.

What I found fascinating is that this addresses the same issue: how to objectively determine ratios of fat and meat. I hope the folks at Stemmer read the report because it goes in to much detail about how the appearance of the meat can vary.

Variation in the object being inspected is of course a big challenge for the development of automated inspection systems. So big in fact that I think I’ll return to it very soon. Until then, how’s that bacon sandwich looking?

No comments: