A style classification algorithm for ink painting based on deep network feature aggregation

Abstract

This study proposes a style classification algorithm for ink painting works based on deep network feature aggregation. Introduce the deep classification regression tree feature aggregation method, use Bayes discriminant criterion, obtain probability density function, and extract the style features of ink painting works. Based on this, a sample dataset for ink painting style classification is constructed, and the Apriori association rule is used to use the corresponding ink painting style category as the final output result of the classifier to complete the classification. The experimental results show that the research method has good performance in extracting style features of ink painting, with short time consumption, low classification error, and high fault tolerance.