Surface Defect Detection on Fixed Pattern Ceramic Tiles

F. Acebrón, F. López, J.M. Valiente, and J. R. Navarro (Spain)

Keywords

Quality control, tile classification, image registration, surface inspection.

Abstract

Automatic and reliable surface defect detection is an important task in the ceramic tile industry, which has not yet been resolved satisfactorily. This paper introduces a detection method for ceramic tiles fixed patterns, where the production process ensures that the same decorative surface is reproduced on all the pieces. This method is based on ‘compare with a reference’ techniques. A tile without defects is used as a reference to find defects on the inspected pieces. The method is divided into four phases: registration, difference map computation, feature extraction and filtering, and classification. The registration phase is necessary to spatially align both tiles in order to be able to compare them. An accurate registration method based on edge detection techniques is presented. The feature extraction is obtained from a difference map computed using spatial and perceptual colour comparison techniques. This difference map passes a filtering stage to eliminate tiny objects resulting from noise induced at the acquisition and registration stages. Finally, a rule-based classifier is presented. This classifier detects the defects, but does not quantify them. This is useful for detecting faulty series of pieces at the intermediate stages of production lines.

Important Links:



Go Back