The Research And Implementation Of The Tangka Image Retrieval System In XiXia Khara-Khoto

In recent years,the research of Xixia Studies have been continuously enriched.As a part of Xixia Studies,Khara-Khoto Thangka image also has important research value.The development of computer vision has provided technical support for Thangka image retrieval.The development of the Xixia Thangka image retrieval system will be more conducive to the inheritance and development of traditional excellent culture.This paper takes Khara-Khoto Thangka image as the main research object,and divides the image library into four categories:Buddha statue,Bodhisattva,Dependent and Zuer.This paper completes the research of image retrieval algorithm and image retrieval system implementation.According to the characteristics of Thangka images,this paper applies convolutional neural network model and feature extraction method based on RCF edge detection model to extract Tangka image features,and compares the retrieval results obtained by the two methods.Through analysis of the experimental results,it is concluded that the feature extraction method based on convolutional neural network has higher precision,recall,average retrieval accuracy and retrieval efficiency than the feature extraction method based on RCF edge detection.This paper has completed the construction of the Khara-Khoto Thangka image digital retrieval research platform.The system function is mainly composed of user application module,image retrieval module,administrator module,and image data management module.The main contributions of this paper are as follows:1?Applying the pre-trained VGG-16 model to extract the deep features of the Thangka image and achieving more rich image representation and more suitable for the complex characteristics of Thangka image texture information.The retrieval accuracy is higher.2?The edge image obtained by the RCF edge detection model is more clear and complete and the HOG algorithm can extract more effective image contour information.3?The system uses windows as the development platform,SSH as the development framework,Java and Python as the development language,and open source Mysql language to build the relational database and developes the Khara-Khoto Thangka image retrieval system.