Development of a machine vision system for belt inspection based on video images.
Machine vision system is largely based on edge detection, i.e. the detection of objects that are classified as a sort of disturbance, clearly visible in the uniform background. Therefore, in case of a machine vision system the ability to name the damage is of lesser importance than to detect it with great accuracy. This is why the main focus was to design a methodology for damage identification, i.e. the ability to detect any damage (edge non-linearity) visible on the belt’s surface in special laboratory conditions on the test rig.
The system was tested on a belt conveyor compatible with 400 mm belts. The system is operational, it measures the belt’s width, detects events by measuring their length. It is nevertheless necessary to purchase two or three additional cameras and to upgrade the software. This will allow the system to cover the belt’s complete width.
The machine vision system mounted on the test rig
Some efforts were also made to detect events with the use of a high resolution thermal vision camera. The effects are not satisfactory enough to enable the usage of thermal photography for belt damage detection, but are a promising starting point for further research. The result of this research is Martyna Konieczna’s engineer disertation titled “Thermal Imaging Methods Used for Estimating Wear Degree of Belt Conveyor Elements”, Wrocław 2013.