Shape-based retrieval of industrial surface defects using angular radius Fourier descriptor

Current industrial surface inspection systems are capable of detecting various defects and producing images of them. The defect images are collected into large image databases. Effective retrieval methods are necessary to analyse the defects stored in the database. In this work, a novel shape descriptor for surface defect image retrieval is presented. The descriptor is based on the Fourier transform of the defect boundary. The proposed method, angular radius Fourier descriptor, combines the boundary function with the directional angle of the boundary line. The proposed method outperforms ordinary Fourier descriptors in the retrieval of paper defect shapes without increasing computational cost.