EyeVision detects scratches or other damages on surfaces with an easy-to-use and graphically configurable Deep Learning Homogeneity Inspector. But not all scratches are always immediately visible for the camera.
Making Scratches visible with Polarizing Filter
Plastics or metals have polarizing characteristics. This characteristic can also be useful for the user and with a polarizing filter on the camera lens the scratches on the surface can be made visible. Additionally, reflections can be reduced and edges can be made more defined. For that color and grey scale cameras can be used.
The articles picture shows an allegedly smooth, black surface of a plastic housing, captured with a Dino-Lite microscope with a polarizing filter. Here an abundance of invisible scratches are made very clearly visible.
Deep Learning Homogeneity Inspector
The EyeVision software can detect damages on surfaces with different methods. Either with the classical image processing method, or with the new Deep Learning algorithms, which are much more suitable for such an application. For the EyeVision Deep Learning Homogeneity Inspector the user does not need a formalized specification or even a threshold area. Completely without parametrization the good parts are taught-in with the One-Class-Classifier and therefore damaged parts can be detected.
The EVT Homogeneity Inspector bundles complex functionalities into a easy-to-use EyeVision command, where the user has a simple graphic tool available and where only a few parameter have to be adjusted. Therefore the training process is very easy-
This works for example on surface inspections of metal, textiles, plastic or fiber board. But the described method for automated quality control works on many surfaces and different data types.