Deep Learning tool opens up new possibilities for materials inspection
Application and Solution
Classical image processing quickly reaches its limits, especially in the area of surface inspection. Despite many options for image pre-processing and image filtering, some defects are difficult to separate from the background, which often results in high pseudo-rejects or undetected flaws. With the help of senswork's Neuralzye® software, an AI inspection of transparent belt material has now been implemented that allows the detection of bubbles and inclusions. Through the additional use of so-called transmitted light deflectometry, the defects in the material can be made visible in the first place.
Diffraction effects caused by a strip of transmitted light create a signal in the line camera image at the location of the material damage that is clearly visible to the human eye. However, detection using standard algorithms is difficult because the background of the image corresponds to a sinusoidal pattern. Based on this image information, Neuralyze detects bubbles, holes and other defects in the material with a high degree of reliability.
Learn more about Neuralyze® image processing software.
- Detection of bubbles or holes
- Detection of inclusions or other defects