AUTOMATIC DETECTION OF SURFACE DEFECTS OF CERAMIC TILES

Abstract

Ceramic tiles are one of the most demanded finishing materials, and the areas of their use are constantly expanding. The production of ceramic tiles is quite well automated, but the control of manufactured tiles for defects is usually not automated, that limits its speed and does not guarantee the necessary quality. The algorithm for detecting the main surface defects of ceramic tiles, namely, mechanical (scratches, cracks, etc.), geometric (chips on the corners, gouges or protrusions on the edges), color defects (blobs, spots, etc.) is proposed in the paper. It based on digital image processing of tiles and was implemented using the free library of algorithmic primitives OpenCV. The algorithm does not require the presentation of reference tiles. At the same time, a high speed of classifying monochrome tiles into fresh and defective ones is ensured and allows to organize the automatic control of plain tiles on the conveyor in real time with the probability of correct detections 97 %.