Ancient Book Paper Inspection Based On The Features Of Book Paper Fiber Morphology

Ancient books have important artistic value and document value,paper inspection of ancient books is an important research direction of ancient book inspection.The disadvantage of the traditional ancient book paper inspection method is that the ancient book cannot be automatically inspected and the ancient book fiber stack is disorderly,which makes the inspection difficult.In order to solve the above problems,this article will conduct research from two aspects: classification of ancient books and paper dating analysis of ancient books.The main tasks are as follows:Aiming at the shortcomings of traditional ancient book paper classification methods,a paper classification method based on the fusion of multi-scale paper fiber morphology features is proposed.On the basis of analyzing the paper fiber morphology characteristics.Firstly,use convolutional neural network to classify ancient books,which overcomes the shortcomings of being unable to automatically classify ancient books and papers.At the same time,in order to solve the problem of difficult classification of ancient books caused by complex fibers,the multi-scale paper fiber morphology feature fusion method was used to obtain more discriminating fiber morphology features.Then,in order to eliminate the interference of feature redundancy between fiber morphology features of different scales on paper classification,a feature fusion and feature selection algorithm based on mutual information is designed.This paper selects 61 ancient books as the research object,compared with the traditional paper classification method of ancient books,the method proposed in this paper achieves a satisfactory effect of paper classification of ancient books.In order to overcome the shortcomings of traditional ancient book paper dating analysis methods,a recurrent neural network based ancient book paper dating analysis method is proposed.First,the cyclic neural network is used to study the relationship between the paper fiber morphology and the production year of ancient books,which overcomes the difficulty of automatically analyzing the production year of ancient books.At the same time,in order to eliminate the interference of paper fiber containing broken fibers and fiber diversity on paper dating analysis,a paper dating regression model based on mixed attention mechanism is proposed.The mixed attention mechanism can not only locate the complete fiber area,but also detect the complete fiber area with unique discrimination,which solves the problem of difficulty in ancient books dating analysis.Then,in order to eliminate the interference of non-discriminatory fibers on the analysis of the production years of ancient books,a new loss function was designed to enhance the robustness of the model.This paper selects 38 domestic books published from 1888 to 2000,compared with the traditional paper dating analysis method of ancient books,the method proposed in this paper achieves a good effect of predicting the production year of ancient books.