Research On Aided Identification Of Ancient Paintings Based On Artificial Intelligence

Traditional Chinese painting is not only a form of ancient painting art,but also a form of expression of Chinese traditional culture.The identification of traditional Chinese paintings is conducive to the appreciation and protection of Chinese traditional culture.But there are two problems:(1)In the Internet age,digital Chinese painting has gradually become popular due to its easy storage and fast retrieval speed.In the task of authenticity identification,experts cannot quickly locate painters in different styles of unsigned digital Chinese painting images,which affects the efficiency of identification.Based on this,the intelligent classification of Chinese painting painters,as an important link in the auxiliary identification of digital Chinese paintings,is essential to improve the efficiency of auxiliary identification of digital Chinese paintings.In recent years,as an important technology for realizing artificial intelligence,deep learning has been applied to all walks of life.How to use the current deep learning technology to improve the accuracy and automation of the intelligent classification of Chinese painting painters has become a key issue.(2)With the development of the times,the fake forms of paintings are no longer just traditional methods such as copying,copying,and depicting.The use of a new generation of information technology to imitate the painter's style of painting makes it difficult to identify the authenticity of the generated painting.However,the traditional identification method that mainly relies on expert experience is subjective and easy to be interfered,which affects the accuracy of identification.Therefore,in the digital appraisal of traditional Chinese painting,an objective and quantifiable reference index is particularly necessary.The proposal of deep learning technology has promoted the development of artificial intelligence and brought opportunities for the authentication of digital Chinese paintings.How to apply this technology to the task of assisting the authentication of the authenticity of digital Chinese paintings,aiming to improve the efficiency and accuracy of the authentication,has become another urgent problem that needs to be solved.This paper analyzes and studies the above-mentioned problems,and the main work is divided into the following two aspects:(1)In the task of authenticity identification,experts face the problem that different styles of unsigned digital Chinese painting images cannot quickly locate the painter category.Based on multi-feature fusion and deep semantic expression of features,this paper proposes an intelligent classification algorithm for Chinese painting painters based on a multi-branch attention mechanism network.First,the algorithm uses contrast statistical features to eliminate digital Chinese painting image blocks with less information;secondly,it uses Chinese painting style characteristics to pre-generate digital Chinese painting features;then,an intelligent classification network based on a multi-branch attention mechanism is designed to complete the Chinese painting Feature fusion and classification.Finally,the experimental results show that compared with traditional mainstream classification methods,this algorithm has advantages in feature extraction and classification accuracy.(2)Based on the intelligent classification of Chinese painting painters,this thesis conducts further research on the authenticity of the painters' paintings.Aiming at the problem that the traditional identification method based on expert experience lacks objective basis,this paper proposes an auxiliary identification method for the authenticity of digital Chinese painting based on the twin network based on the deep learning model.First,the adversarial generation network is used to learn the artist's artistic style and generate digital fakes of the corresponding painter.Secondly,for the generated fake works,a deep learning-based digital Chinese painting authenticity auxiliary identification network is proposed.Then,visualize the feature area of the painter's image.Finally,an auxiliary authentication system was developed based on the web.Through the analysis of experimental results,the average identification accuracy of this method reached 0.833,indicating that it has a good sensitivity to digital forgery.