| Summary: | Based on the content analysis method, this paper first uses the PCA technique to disassemble and transform the base color information to extract multiple feature values. Secondly, the sparse autoencoder is used to adjust the feature parameters by training, and the activation function pair is used to calculate the confidence to complete the effective classification. Finally, the SVM algorithm is introduced to optimize the color information parameters, and the corresponding color label values are calculated by constructing a decision function to accurately set the pixel positions. The results show that the color difference value of Chinese oil painting image keeps between 0.5nbs -1.5nbs, the standard deviation value of saturation is 165.55, and the resolution is 610ppi, which indicates that the content-based analysis method can better coordinate and integrate the relationship between various color structures.
|
|---|