基于图像处理技术的建筑物沉降研究  

Research on Building Settlement Based on Image Processing Technology

作　　者：张锐[1] 蒋慧莹 ZHANG Rui;JIANG Hui-ying(college of Automation,Harbin University of Science and Technology,Harbin Heilongjiang 150080,China)

机构地区：[1]哈尔滨理工大学自动化学院,黑龙江哈尔滨150080

出　　处：《计算机仿真》2022年第9期247-251,共5页Computer Simulation

摘　　要：建筑物沉降观测是保证建筑物安全的重要内容,为解决传统的建筑物变形监测方法成本高、安装固定难的问题,提出了目标建筑物摄影观测的方法,对各期监测图像相同监测点在垂直方向产生的建筑物沉降量进行分析计算。其关键在于首先确定相机单位像素的物理尺寸,并对拍摄图像进行相机标定,利用得到的内参和外参矩阵纠正拍照过程中产生的桶型和枕型畸变,其次对标定后的图像进行超分辨率重建,通过训练卷积神经网络模型更新结构参数,最终重建后的图像的峰值信噪比达33.00db,最后对拍摄图像进行SIFT特征点提取和匹配,计算各期图像像素间的差值令沉降量精确至毫米级,达到国家衡量建筑物沉降基准。In order to solve the problems of high cost and difficult installation and fixation of traditional building deformation monitoring methods, a method of photogrammetric observation of target buildings is proposed. The settlement of buildings in the vertical direction generated by the same monitoring points in each period of monitoring images was analyzed and calculated. Firstly, the physical size of the unit pixel of the camera was determined, and the camera was calibrated for the captured image. The barrel type and pillow type distortion generated in the process of photographing were corrected by the obtained internal and external parameter matrices. Secondly, the calibrated image was super-resolution reconstructed, and the structural parameters were updated by training the convolution neural network model. The peak signal-to-noise ratio of the reconstructed image is 33.00 db. Finally, SIFT feature points were extracted and matched to calculate the difference between the pixels of each phase of the image, so that the settlement was accurate to millimeter level, which reached the national standard for building settlement.

关 键 词：建筑物沉降 相机标定 成像模型 超分辨率重建 特征提取 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]