基于亚像素圆心检测法的变形监测技术

TECHNIQUE FOR DEFORMATION MONITORING BASED ON SUBPIXEL CENTRE DETECTION ALGORITHM

通过分析当前应用于变形监测的图像处理技术,基于传统的灰度重心法和圆边缘检测法,提出改进的有标点中心定位法——亚像素圆心检测法。为验证其精度,设计对比试验,从标点尺寸、拍摄距离、光源强弱和标点运动状态4个方面,对比分析改进算法相对于传统算法的优越性,并得到改进算法的理论精度。同时,通过研究标点位移的特点,建立基于亚像素圆心检测法的变形监测技术,并设计试验,从拍摄距离、光源强弱、标点运动状态3个方面分析该技术的理论可行性。试验结果表明:有标点亚像素圆心检测法精度较高,在50 m拍摄距离内精度达1 mm(15 cm直径圆);基于该算法的变形监测技术对标点坐标变化敏感,在40 m拍摄距离内,识别误差控制在1.0 mm范围内,在25 m拍摄距离内,误差严格控制在0.5 mm内,可行性强。依托江西省石城—吉安(石吉)线B8标李家寨高边坡,进行现场试验,监测结果显示,监测误差与理论研究结果相符,表明新型监测技术具有较好的工程适用性。经理论研究与实践证明,该算法监测精度高、可行性强、工程适用性强、成本较低、可实时反馈监测信息。

By analyzing the present image processing techniques applied to deformation monitoring, the advanced target method--subpixel centre detection algorithm is proposed based on the traditional gray weighted detection algorithm and detection algorithm of circle edge. The accuracy test is designed to research its advantages comparing with the traditional algorithms through 4 aspects： target size, camera distance, light intensity, state of target motion. In return, we can get the theoretic accuracy of the method. Based on the features of mark displacement, the new technique of deformation monitoring is put forward. And the applicability test is carried out considering 3 aspects： camera distance, light intensity, state of mark motion. The results indicate that： （1） The subpixel centre detection algorithm with targets can get the higher identification precision with less than 1 mm in shooting distance of 50 m（circle of diameter 15 cm）. （2） It is feasible with its sensitivity to catch the markdisplacement. When the shooting distance is less than 40 m, the identification error could be controlled to 1.0 mm. And when the shooting distance is less than 25 m, the identification error could be strictly controlled to 0.5 mm. In addition, according to Shicheng--Ji＇an highway project in Jiangxi province, the new technique is applied to detect the slope deformation, which testifies its high engineering applicability. The theoretical study and tests confirm the higher accuracy rate, technical feasibility, engineering applicability, and feedback ability in time of the new technique based on the advanced algorithm.