Ancient Chinese Character Image Retrieval Based On The Fusion Feature Of DHFS And DWFnet

Image retrieval of ancient Chinese characters can assist researchers in tracing similar glyphs in the research process,which is an effective tool for related research of ancient Chinese characters.However,due to the aging of ancient documents and the complex glyph structure and variety of Chinese characters in ancient books,the accuracy of image retrieval of Chinese characters in ancient books is affected,and the existing text image retrieval and recognition technology is difficult to achieve the ideal retrieval effect for the ancient Chinese image.Therefore,aiming at the characteristics of ancient Chinese character images and the shortcomings of traditional Chinese character image retrieval techniques,an image retrieval method based on the fusion feature of dual hesitant fuzzy sets and discrete wavelet fusion network is proposed.By introducing DHFS,DWFnet and canonical correlation analysis,the image retrieval features of ancient Chinese characters were extracted,which integrated with structural feature and deep feature,and the image retrieval model of Chinese characters in ancient books was constructed.The main work includes:(1)Extraction of structural feature of ancient Chinese character imagesDual hesitation fuzzy sets that can express uncertain information more comprehensively are employed in the feature extraction process of directional line elements.The multi-attribute evaluation index of adjacent grids for the current grid and its corresponding membership and non-membership functions are established,and the weight of each attribute is calculated by the dual hesitation fuzzy entropy,such that the proposed features can fully reflect the topological structure of ancient Chinese characters.The experimental results show that the feature of dual hesitating fuzzy direction line element is better than other artificial features to describe the structural features of Chinese characters in ancient books.(2)Extraction of deep feature of ancient Chinese character imagesThe image feature extraction method of ancient Chinese character image based on discrete wavelet fusion network,which integrates the low-level features and the high-level features of the convolutional neural network effectively,and the fusion feature map is compressed into feature vectors by Spatial Pyramid Elastic Pooling.This method avoids the problem that the traditional convolutional neural network only uses the high-level features to classify and ignores the detail features of the low-level network.Experimental results show that DWFnet can extract multi-layer fusion features with more discriminant information,and improve the retrieval performance of ancient Chinese character images.(3)The ancient Chinese image retrieval model based on the fusion feature of DHFS and DWFnetThe ancient Chinese image retrieval model based on the fusion feature of DHFS and DWFnet is proposed.The dimension of DWFnet feature is reduced by principal component analysis.Dual hesitating fuzzy direction line element features and DWFnet features were fused by canonical correlation analysis to establish the ancient Chinese character image retrieval model based on multiple fusion features.The experimental results show that the fusion feature has stronger representation ability than the single feature,and is more suitable for the image retrieval of Chinese characters in ancient books.