历史建筑内墙表皮病害非接触式检测方法研究与应用

曾恬

摘 要

历史建筑极具重要性和特殊性，是文化和技术的实物载体。随着时代的发展进步，历史建筑历经风霜，内墙表皮病害问题突出，影响美观甚至危害结构安全。在历史建筑的科学保护前提下，传统的检测方法如目测法和接触式检测方法已无法符合历史建筑的保护要求。因此，在工程实际中需要新方法新技术来实现对历史建筑内墙表皮病害的检测。

本文针对历史建筑内墙表皮常见的危害性较大的潮湿渗水病害和空鼓脱落病害问题进行了非接触式检测方法的研究。针对历史建筑墙体表面的潮湿渗水病害,首先对获取的墙面图像进行色度空间转换和滤波的处理，并且比较各色度空间的直方图和滤波处理算法的效果，其次运用OTSU算法得出最佳的阈值并对图像进行二值化处理，最后提取病害区域，达到了潮湿渗水病害检测的目的。针对历史建筑的墙体表面的空鼓脱落病害，首先根据点云平面模型拟合的基本原理以及传统的RANSAC算法的缺陷，给出了优化RANSAC算法计算点云平面模型的拟合方法，并通过该方法提取出建筑墙面的平整度不均匀的部分，从而实现了检测墙面空鼓脱落病害的目的。

本文所研究的病害非接触式检测方法在历史建筑病害系统上进行了集成，并在巴公房子的实例中进行了检测应用。证明本文的方法能够很好地识别出墙面的潮湿渗水病害和空鼓脱落病害，计算出病害的面积以及相应的修缮建议，能够为后续的历史建筑病害的溯源提供借鉴意义。

关键词： 历史建筑；病害检测；非接触式检测；图像分割；点云平面拟合

Abstract

Historic buildings are of great importance and particularity, and are physical carriers of culture and technology. With the development and progress of the times, historical buildings have experienced wind and frost, and the problem of internal wall skin diseases is prominent, which affects the appearance and even endangers the structural safety. Under the premise of scientific protection of historical buildings, traditional detection methods such as visual inspection and contact detection methods can no longer meet the protection requirements of historical buildings. Therefore, new methods and new technologies are needed in engineering practice to realize the detection of skin diseases on the inner walls of historical buildings.

In this thesis, a non-contact detection method is studied for the common and more harmful moisture and water seepage diseases and hollow shedding diseases of the inner wall of historical buildings. Aiming at the moisture and water seepage diseases on the wall surface of historical buildings, firstly, the acquired wall images are processed by chromaticity space conversion and filtering, and the histograms of each chromaticity space and the effects of filtering processing algorithms are compared. The optimal threshold value is obtained and the image is binarized, and finally the diseased area is extracted, which achieves the purpose of detection of wet and seepage diseases. Aiming at the hollowing off of the wall surface of historical buildings, firstly, according to the basic principle of point cloud plane model fitting and the defects of the traditional RANSAC algorithm, a fitting method of optimizing the RANSAC algorithm to calculate the point cloud plane model is given. The method extracts the uneven flatness of the building wall, so as to achieve the purpose of detecting the hollowing and falling off disease of the wall.

The disease non-contact detection method studied in this thesis is integrated on the historical building disease system, and the detection application is carried out in the example of the Bagong house. It is proved that the method in this thesis can well identify the damp water seepage disease and hollow shedding disease on the wall, calculate the area of the disease and the corresponding repair suggestions, which can provide reference for the subsequent traceability of historical building diseases.

Key words: Historic buildings; disease detection; non-contact detection; image segmentation; point cloud plane fitting