Possible sources of both system errors and random errors for digital image processing system have been discussed. Calibrations have been made to get ride of system errors such as lens distortion. As for random errors, necessary improvements for experimental environment and data processing have been made. After all these measures taken, other error sources remained are not quantified separately, instead, accuracy of the entire system is examined. Capability of using this system for strain measurement in triaxial tests is illustrated throught comparisons of strain results obtained from both this system and conventional measurement techniques. Finally, the accuracy of this system is quantified through comparison of strain results obtained by both image processing technique and a static strain tester, and the statistical accuracy value is given out to be 4×10-5.%针对已研发的土工三轴试验试样全表面变形数字图像测量系统,从系统误差和随机误差两方面分析各可能的误差来源,并从系统误差修正和试验环境改善两方面着手消除或减小误差.以具代表性的饱和无黏性土三轴排水剪切试验为例,通过由图像测量系统所测得的试样变形结果与传统手段观测结果的对比,讨论图像测量系统用于三轴试验的合理性.进一步给出一种应变测量精度检定方法:以静态应变测试仪所测得的试样应变为“真值”,将图像测量系统所测得的应变值与应变“真值”作比较,通过大量试验数据,统计出本图像测量系统的误差,进而给定了图像测量系统的实际测量精度为4×10-5.同时还比较体积应变的测量结果.
展开▼
