Automated defect detection in case of reconstruction of archive documentary photographs

The paper developed a method for detecting defects in the archival documentary photographs by texture analysis and allocation of local descriptors in images. The algorithm developed by the modified method SURF allows selecting different types of local features with image pre-processing power characteristics Low’s. The proposed approach allows us to find points of interests on the different types of textures and analyze defects on the surface of the images as a type of texture, such as edges, spots, ripples, waves, etc.