PDF全文阅读

彭方圆

 

（ 北京城建房地产开发有限公司 ，北京 100081)

摘   要 ：为了能够精准识别出装配式建筑预制梁下外墙板拼缝渗水状态，评估建筑使用质量，提出建筑预制梁下外墙板拼缝渗水状态红外图像识别方法。应用高斯低通滤波方法和基于控制点的几何校正模型对外墙板拼缝红外图像进行降噪与校正处理；应用基于热源模糊聚类方法对处理后的红外图形进行聚类分割；根据图像分割结果构建基于 BiFPN 改进的 YOLO v3 网络模型，提取外墙板拼缝图像渗水特征、聚类分割后的外墙板拼缝渗水红外图像特征 ，精准识别外墙板拼缝渗水红外图像。结果表明：应用本文方法的峰值信噪比均在52. 6 以上，交互比最高值为 0. 986 ;  红外图像预处理效果良好，识别精度较高，有效满足建筑红外检测标准要求，为评估建筑的使用质量提供可靠依据。

关键词：装配式建筑 ；预制梁 ；外墙板 ；拼缝渗水状态 ；红外图像识别 ；聚类分割

中图分类号：TP69           

文献标志码：A            

文章编号： 1005- 8249   (2024)  04- 0111- 06 

DOI:10. 19860/j.cnki.issn1005 - 8249.2024.04.020

 

PENG Fangyuan

(Bucg Real Estate Co. ，Ltd. ，Beijing 100081 ,   China)

Abstract: In order to accurately identify the water seepage state of the joint of the outer wall panel under the prefabricated beam of the prefabricated building and evaluate the service quality of the building, an infrared image recognition method for the water seepage state of the joint of the outer wall panel under the prefabricated beam of the building was proposed. Apply Gaussian low-pass filtering method and geometric correction model based on control points to denoise and correct infrared images of external wall panel seams; Apply the fuzzy clustering method based on heat source to cluster and segment the processed infrared images; Construct a YOLOv3 network model based on BiFPN improvement based on image segmentation results, extract water seepage features from exterior wall panel joint images, cluster the segmented infrared image features of exterior wall panel joint water seepage, and accurately identify the infrared image of exterior wall panel joint water seepage. The results show that the peak signal-to-noise ratio of the proposed method is above 52.6, and the highest interaction ratio is 0.986. The infrared image preprocessing effect is good, the recognition accuracy is high, and it effectively meets the requirements of building infrared detection standards, providing a reliable basis for evaluating the quality of building use.

Keywords: prefabricated building; precast beam; outside the wall; splice seam seepage state; infrared image recognition; clustering segmentation

 

作者简介： 彭方圆  (1979—) ,  男 ，本科 ，工程师 ，研究方向：装配式建筑施工。

收稿日期：2023- 02- 17