Research On Computer Vision Detection Of The Ceramic Tiles' Color Difference

Ceramic tile is a kind of commonly used building material.With the improvement of living standard,the requirement for it's quality also becomes more and more serious.Due to the complexity of the producion,the ceramic products have the non-homogeneous problems in the color part.Aim at the color-difference detection problem exists in the ceramic field,a new color-difference online measuring method basing on the machine vision is proved in this paper.It can solve the problems to a great extent which occur during the mass production of tile,such as the slow-speed of manual detection, large labour quantity, harsh environment, subjective factors are great influence to the result and so on.This subject is developed by College of Precision Instrument Tianjin University and South China Institute of Precision Manufacture. This paper contact disciplines trends and the needs of practical application, studies the linear CCD large area scanning technology,the image preprocessing and analysing;then the system of ceramic tile's color different deteciotn is designed.The system which integrates application optics technology, computer technology, digital image processing technology,accurate measuring technique is on the basis of machine vision.Fistly,the possibility of image collecting system's application are analyzed from the view of hardware in this paper.The properties of illumination devices,CCD,cameras,image collecting card and IO card and how to choose them are discussed. Secondly,several image processing methods are analysised and compared in designs and experiments,then the methods fit for the research are used, such as conversion between color module,image denoising,segmentation and so on.Considering the particular situation of the research,HIS color module,median filter and threshold model are choosed.Finally,the classification of ceramic tile is determined by the method based on histogram distribution.Lots of experiments did in laboratory enviroment verificated the feasibility of this method which can be used to classficate ceramic tile dynamicly.It has good repeatability and stability and three critical technology is obtained.