建筑构件缺损部位修复方法分析

Analysis of Repair Methods for Defective Parts of Building Components

摘要:

文章针对建筑构件缺损部位,提出了一种融合计算机视觉、多模型识别、深度学习匹配和3D打印的智能修复方法。该方法通过无人机多角度采集缺损构件图像,利用运动结构恢复(Structure from Motion,SFM)算法建立三维模型并提取轮廓线,借助训练好的模型识别饰面类型,再结合深度学习算法在构件数据库中匹配匹配度最高的修复样例,生成缺损部位三维模型后通过3D打印制作修复部件,最终完成缺损构件的安装与饰面施工。研究表明,该智能修复方法能够有效提升修复构件与原构件的造型风格一致性及饰面装饰匹配度,具备良好的自动化水平与工程适应性,对推动建筑修复向智能化、精细化方向发展具有重要意义。

Abstract:

This article proposes an intelligent repair method that integrates computer vision, multi model recognition, deep learning matching, and 3D printing for damaged parts of building components. This method collects images of damaged components from multiple angles using drones, establishes a 3D model using SFM algorithm and extracts contour lines, identifies the type of decorative surface with the trained model, and then matches the repair sample with the highest matching degree in the component database using deep learning algorithm. After generating a 3D model of the damaged area, the repaired parts are made through 3D printing, and finally the installation and decorative surface construction of the damaged components are completed. Research has shown that this intelligent repair method can effectively improve the consistency of the styling and decorative matching between the repaired and original components, and has good automation and engineering adaptability. It is of great significance for promoting the development of intelligent and refined building repair.