Failure detection and isolation in ceramic tile edges based on contour descriptor analysis

Today’s industry is based on the principle of automatics. In order to increase production quality and improve production yield that reduces production costs, there is tendency to automate every phase in the production line and if possible to replace the human resource entirely. Ceramic tile industry is not an exception. In the ceramic tile production line almost every phase is more or less automated. The only part which is least automated is visual inspection and classification of a ceramic tile. In this paper we present a method for visual inspection of ceramic tile edges, their failure detection and isolation using a machine vision system. For edges inspection, detection and failure isolation we use a method based on contour tracing and description of tile edges. The method developed generates a contour reference descriptor based on directional contour tracing, determines tile orientation for referenced preprocessing and localizes and isolates the failure. The method is tested and results are given for several types and sizes of ceramic tiles.