Tile Identification And Classification Based On Fusion Of Multi-channel Features

Tiles are regarded as a necessity for people's lives.With the continuous changes of social needs,the types of ceramic tiles have gradually increased.The manufacturers are constantly improving the level of science and technology and smart manufacturing to meet the needs of visual identification and intelligent control in tile manufacturing.A method of classification and identification based on multi-channel feature fusion is presented with the visuals of tile in this article.Inspired by the biological vision system processing mechanism,a new method is proposed in this paper for tile classification in which a multi-channel feature fusion algorithm is used.In order to achieve a better tile classification effect,according to the multi-channel processing mechanism,RGB is converted into the HSI color space which is more similar to representation in the human visual system.Then the model of HSI is converted into three channels of H,S,I.Two types of features were extracted and established classifiers in this article.The features of texture and HU moment invariant are extracted in each channel during this process.The multi-channel features are fused with a new algorithm.Finally,SVM is implemented and used as a classifier for classification.In the production process,the tile images are blurred for wobble or other factorsThe final recognition effect is affected.The histogram equalization is adopted for each channel to enhance the texture of the tile in this article.In order to obtain higher accuracy,this paper uses different kernel functions to perform experiments and parameter optimization on SVM.The result is shown that the performance of RBF kernel function is the best.The recognition rate of multi-channel feature fusion algorithm is 20%higher than that of single color or texture method.The fusion recognition rate reached 94.55%.The research can be applied to the production line of tiles,replacing human with machines to achieve intelligent control.It not only can reduce the labor force,but also can improve the work efficiency and has certain application value.