Research And Realization Of The Intelligent Detection And Classification System For Ceramic Tile

Aiming at the problems of the manual detection in the tile quality detection by which caused the classification accuracy and classification efficiency of low-status,and the singleness of feature parameter criterion in the current automatical detection technology.In this text,a detection system based on machine vision and focused on BP neural network algorithm was designed.The system consisits of light source,image capture devices,image pre-processing and intelligent judgments modules.In the system,the tile image was acquired by using CCD camera,the filtering and edge operator methods were used to deal with it,which aims to complete the removing of environment light,the smoothness of the image and the intensifying of the image.When the neural network was identified to classify tile image,whose color features,texture features and shape features that the three composite indicators was regarded as input vector.The expression and calculation method of each eigenvector was analyzed,the color feature and shape feature of tiles were extracted by using MATLAB software and the grain feature was calculated by using VC platform.At the same time,the three characteristics vector was to carry out comprehensive training,the number of hidden layer units was confirmed in the tile image detection classification.Then several improved BP neural network algorithms were compared,and the optimal BP neural network model was confirmed.The three feature parameters was regarded as the input vectors to a BP neural network,then the tile was classified by the judgement of BP neural network to the tile's quality,the intelligent detection and classification system for ceramic tile was realized based on neural network.At last,the original data of tiles was predicted,to which combined with regression analysis and gray theory.The neural network included original data and the prediction data was set up,which aims to classify and judge tiles.Then the study example demonstrated the comprehensive detection of tile's quality was completed fast and accurately by tile quality dectection.