Large Flatbread Inspection

A vision system is required to inspect large sheets of a hard flatbread product, post baking and before it is cut into the retail size. The product is on a wide conveyor so a series of cameras are required to monitor several conditions.

The Scope

The vision system’s job is to inspect the whole surface area of the ‘biscuit’ and look for the following flaws:

  1. Sheet is complete
  2. No breaks, cracks, holes
  3. No burn (area size to be determined)

If a defect is detected, a signal will be generated to remove the product from the belt.

This proposal only covers the inspection of the above criteria however, future requirements may include the detection and measurement of topping such as cereals and seeds.

The Challenges

The product is very large as it is cut into much smaller sizes after it has been inspected. This requires multiple cameras to work in unison to inspect the whole sheet in one go.

  • The sheets are approximately 2150mm across the width of the conveyor
  • Sheets are 900mm in length along the direction of travel.
  • Constant average colour and texture but colour may vary
  • Conveyor moves at 15 to 20 m/minute

The Results

This is a high performance system that has to inspect a very large surface area in real time. In order to do this the images from each of five cameras are stitched together to create one large image of the ‘biscuit’ before the inspection is carried out.

Setting up this camera configuration took some special effort, with test cards and tape measures in order to achieve the accurate image stitching.

Eventually, all five cameras were synchronised and a complete 2 metre wide product was imaged.

Some of the production flaws were interesting.

The results are good as this system guarantees 100% inspection, 100% of the time with little or no human intervention. Saving the manufacturer money in terms of electricity from heating expensive to run ovens, ingredients and of course, labour costs.

Reposted from Scorpion Vision Ltd Blog dated 27th November 2019

Scroll to Top