摘要
碑帖承载着丰富的历史文化信息,然而由于自然侵蚀和人为破坏等原因会导致其中语义信息丢失,进而给识别碑帖年代造成困难。为此,本文结合数字人文和深度学习,旨在构建一种高效的碑帖年代识别系统,为碑帖文化研究提供支持。在深度学习领域,卷积神经网络技术已经得到了广泛的应用。本文选用残差网络,构建图像分类模型,使用明、清、民国3个朝代的碑帖图片训练模型,通过图像分类实现对于碑帖年代的识别。本文构建的碑帖年代识别系统能够以较高性能实现对于明、清、民国3朝代碑帖的年代识别,便于研究者更加迅速和准确地识别碑帖年代,充分挖掘其中蕴含的历史文化信息,促进文化遗产的保护与开发。
Stele inscriptions harbor a wealth of historical and cultural information.However,the loss of semantic content due to natural erosion and human-induced damage poses a challenge in accurately determining their age.This paper addresses this issue by integrating digital humanities with deep learning,aiming to establish an efficient system for age identification in stele inscriptions,and to provide support for the study of inscription culture.Leveraging the widespread application of convolutional neural network technology,this study employs residual networks to construct an image classification model,which is subsequently trained on stele images spanning the Ming and Qing Dynasties and the Republican era.The model,through image classification,achieves the goal of age recognition.The resulting system demonstrates high performance in identifying the age of inscriptions from these dynasties,enabling researchers to identify inscription periods swiftly and accurately.This advancement not only helps unlock the historical and cultural information embedded within the inscriptions,but also contributes to the preservation and advancement of cultural heritage.
作者
周知
周群
Zhou Zhi;Zhou Qun(School of Public Administration,Northwest University)
出处
《图书馆杂志》
北大核心
2025年第8期81-89,115,共10页
Library Journal
基金
教育部人文社会科学青年基金项目“用户认知结构视角下人文图像资源标注研究”(项目编号:21YJC870023)的阶段性研究成果之一。
关键词
数字人文
碑帖
图像分类
卷积神经网络
Digital humanities
Stele inscriptions
Image classification
Convolutional neural network
