Research on the Application of Pattern Recognition in the Intelligent Identification of Wallpaper Labeling
In order to realize the intelligent identification of wallpaper labeling, the wallpaper texture and the characteristic of color are comprehensively considered in this paper to rich the pattern feature space. Firstly, the suitable GLCM (gray level-gradient co-occurrence matrix)is constructed to describe the texture feature and extract the feature parameters; And the parameters of color low order matrix are picked up in the RGB color space to constitute mode characteristic vector together. Secondly, to reduce the computation and improve the describing abilities, the Simulated Annealing Algorithm is applied to select feature value from 17 feature parameters. Lastly, the integrated classifier of BP neural network is designed to achieve 94.03% overall recognition rate, which is higher than the traditional one. The experiment results also have shown that this method is effective.
Zhenyu Du and Bin Liu
Y. Q. Wang et al., "Research on the Application of Pattern Recognition in the Intelligent Identification of Wallpaper Labeling", Applied Mechanics and Materials, Vol. 65, pp. 518-523, 2011