On-line automatic detection of foreign bodies in biscuits by infrared thermography and image processing

The on-line implementation of a Thermography Non-Destructive Testing method for the detection of inner contaminants and foreign bodies in biscuits is presented. The procedure exploits an image processing algorithm developed to be applied to the thermograms provided by an infrared camera and it entails an automatic defect detection protocol. It is shown that the proposed procedure allows the automatic on-line detection of most of the defect types considered. To achieve this result, preliminary off-line and on-line tests and their statistical classification analysis were performed. The off-line measurements consisted in the acquisition of temporal sequences of thermograms collected at the exit of the oven during the cooling process of biscuits. The processing is focused on the comparative analysis of the thermal emissivity of both foreign bodies and of biscuits decay curves. Benchmark defects were inserted in the on-line products. The statistical classification analysis was executed on thermograms acquired on these benchmarks and it was aimed to the individuation of an automatic foreign body detection algorithm.