清代神龛构件图像识别及聚类方法研究 #

Image Recognition and Clustering Methods for Shrine Components of Qing Dynasty

摘要:

本文针对大数量且形式趋同的清代神龛构件自主聚类进行研究。通过对图像识别、聚类、机器学习算法的挖掘、改写，以及新算法的编写，利用 Python 编程，本文形成可以将清代神龛构件自主聚类的闭环技术流程。该流程可以辅助完成海量相似构件的聚类，减少人工干预，加快聚类过程。该流程具有推广性，能够扩展应用到其他类似项目，也为计算机技术在文物建筑虚拟复原的具体应用提供了参考。

Abstract:

In this paper, We research, rewrite, and invent image recognition, clustering, and machine learning algorithms with python to propose a closed-loop technology flow that investigates the autonomous clustering of massive Qing dynasty shrine components with convergent forms. This study can assist in clustering a vast amount of similar components, reduce manual intervention, and speed up the clustering process. The output is generalizable and can be extended to other similar projects. It also provides a reference for the application of computer technology in the virtual restoration of heritage buildings.