文章基本信息

标题：An automatic technique for accurate non-contact structural deformation measurements
本地全文：下载
作者：C. Psaltis ; C. Ioannidis
期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2006
卷号：XXXVI Part 5
出版社：Copernicus Publications
摘要：This paper presents an automatic, close-range photogrammetric process developed to measure point-wise 2D deformations of a concrete beam during a loading experiment. Required accuracy was better than 1 mm. A non-contact method is proposed, which is simple to implement and provides almost online results that meet the accuracy requirements fully. This technique requires only one known distance between two signed points, and the use of specifically designed targets. First, a stereopair is acquired by a calibrated digital camera. The camera stays still in its last position, while images are taken periodically during the load epochs. When data collection is completed, the relative orientation of the stereopair is calculated automatically. The image point coordinates are measured using a template matching algorithm, based on normalized cross-correlation, and the homologous points between the two images are located by automatic recognition of the code number of each signed point. Next, 3D point coordinates are calculated by a photogrammetric intersection solution using the known relative orientation parameters in an arbitrary reference system. As all monitored points of the front side of the beam remain practically on the same plane during the load epochs (2D deformations), the image-point coordinates are measured automatically for each epoch, and their 2D-coordinates are calculated. Using the known distance, the scale of the system is defined and all deformations are transformed into the object space. The inner accuracy is determined by comparing consecutive images of each epoch, while the outer accuracy through the comparison of the results with measurements using dial gauge. The results in both cases show that the overall accuracy of the method is approximately 1.5 pixels, or better than 1 mm in object space
关键词：Close-range; Automation; Matching; Recognition; Monitoring; Tracking