Motion-based visual inspection of optically indiscernible defects on the example of hazelnuts

Automatic quality control has long been an integral part of the processing food and agricultural products. Visual inspection offers solutions for many issues in this context can be employed form sensor-based sorting to automatically remove foreign low entities from a product stream. However, these methods are limited defects that made visible by sensor, which usually restricts system appearing on surface. An alternative non-visual solution lies impact-acoustic methods, do not suffer constraint. strongly terms material throughput consequently suitable large scale industrial application. In paper, we present novel approach performs based optically acquired motion data. A high-speed camera captures image sequences test objects during transportation process chute with specific structured The trajectory data is then used classify their behavior. evaluated experimentally example distinguishing defect-free hazelnuts ones insect damage. Results show merely utilizing data, recognition rate up 80% undamaged achieved. major advantage our it integrated systems high applications.