Artificial Vision and Pattern Recognition techniques for the analysis of Textures on ceramic surfaces

摘要

For more than twenty years, computer image analysis has been a valuable instrument in the ceramic production process, especially in the visual inspection phase. These Automated Visual Inspection (AVI) techniques have been developed in two well-established disciplines of the computer world - Artificial Vision and Pattern Recognition. The first deals with models and techniques for emulating the visual behaviour of a biological system, and therefore studies all the computer techniques necessary for the derivation of visual primitives, such as colours, textures, outlines and shapes used in processes for comprehension and recognition. The second uses instruments, and in particular statistical instruments, to analyse, classify and identify patterns and models defined by substantial quantities of measured data, such as the above visual primitives derived from images. Many problems that affect the quality of the ceramic product can be solved thanks to Artificial Vision: the dimensional analysis, for the identification of geometric abnormalities (for example on edges and corners); the analysis of local abnormalities (such as cracks or drops or porosity), the tone and colour analysis and the analysis of textures.