Artwork Image and Retrieval Strategy Based on Feature Fusion and Neural Network

Abstract

To improve the ability of 3D reconstruction and classification of works of art, an image recognition and retrieval strategy based on feature fusion technology is proposed. Firstly, the main methods of visual image feature extraction and color feature decomposition are discussed, and their key technologies and applications are analyzed. Then combined with the characteristics of art images, a multi feature fusion method based on color and texture is proposed. Finally, the results are input into the statistical data of the whole work obtained by neural network. The experimental analysis shows that the fusion analysis of art image feature data uses the advantages and disadvantages of different classifiers and complements them, which effectively provides the accuracy and recall of image recognition. Compared with traditional methods, it can suppress noise better and extract more useful edge information.