Defect or Design? Leveraging the Angle of Opportunity for Detecting Scratches on Brushed Aluminium Surfaces

Design features such as polishing strokes share similarities with defects; this makes defect detection and quality assessment difficult to perform both manually automatically. Human assessors rotate objects probe different incoming illumination angles evaluate the dimension limits samples i.e. decide whether differences between candidates design qualify a defect. This process has poor access quantifiable descriptors needed for automation expose gap in existing evaluation of defects. To integrate notion into automated we propose spatio-temporal image acquisition setup capturing descriptor Angle Opportunity (AoO) which can be used feature image-based classification. The Random Forest approach classified defects an area under ROC-curve 92%.