Description
Multi-sensorial systems combining digital colour linescan cameras with b/w cameras and external sensors into a cluster for the monitoring esthetical appearance, colour shades, local colour defects, physical defects and geometry of continuous webs or single tiles.
The inspection is based on a self-learning, fuzzy colour image comparator technology.

The company is dedicated to the surface inspection of difficult to judge surfaces like multicoloured and textured ceramic tiles, natural looking laminates etc. The sorting systems are designed systems to replace the human sorter. These systems mimic the human colour/texture/aesthetic appearance perception. To achieve this basic human perception and recognition features are applied, to fulfil the task automatically.
The automatic systems are structured into pre-attentive perception simulator and pattern recognition modules.

The systems undergo a learning phase which is restricted to a very small set of samples, because at the beginning of a production there is no statistically sufficiently large set of samples available. To perform on these very poor training samples generalisation rules are applied.

This work does not follow the traditional path of object segmentation, symbolic description, and reasoning about the scene content as our scenes are partially or fully random, complex looking surface patterns which have to be categorised using very poorly understood human visual expertise.

Further Information
www.massen.com

back