针对材料力学实验中的三维变形测量，提出并实现了一种基于双目立体视觉、摄影测量术和数字图像相关法的便携式三维变形测量系统。研究了该系统涉及的双目摄像机标定，图像相关算法，三维重建，以及三维位移、应变计算等关键技术。提出了一种基于摄影测量术的摄像机标定算法，该算法采用10参数镜头畸变模型，不需要高精度标定板即可实现摄像机的高精度标定。利用最小二乘非线性优化算法实现了数字图像的高精度匹配，针对非线性优化初值难求的问题，提出了一种基于种子点的初值计算方法，为非线性优化提供了可靠的初值。最后，介绍了三维重建以及计算三维位移、三维应变的方法。实验结果表明，标定结果的重投影误差为0.03 pixel，图像匹配的误差约为0.02 pixel，静态外形及位移的测量精度为0.05%，应变的测量精度优于0.5%。与传统的测量方法相比，本文提出的系统可以更准确、全面、直观地实现对位移场、应变场的测量。

In order to solve the problem of the three-dimensional deformation measurement in mechanics experiments, a new deformation measurement system based on stereo vision, photogrammetry and a digital image correlation method is proposed and implemented. The key technologies applied in the system are studied, including binocular camera calibration, image correlation algorithms, three-dimensional reconstruction and the calculations of three-dimensional displacement and strain. Firstly, a new camera calibration algorithm based on the photogrammetry is proposed, in which a 10 parameter lens distortion model is adopted to achive a reliable camera calibration result without any accurate calibration pattern. Then,the high precision image correlation is realized by using a least-square nonlinear optimization algorithm.To solve the problem in calculating initial values for the nonlinear optimization, a method based on seed points is proposed to provide a reliable initial value for the least-square nonlinear optimization. Finally, the approaches used for three-dimensional reconstruction and calculating three-dimensional displacement and strain are presented in detail. Experimental results show that the RMS error of the calibration result is 0.03 pixel, the accuracy of image correlation is about 0.02 pixel, and the measurement accuracies for the static profile, displacement and the strain are 0.05% and better than 0.5%,respectively.compared with the traditional measurement method, the system can be more accurate, comprehensive and intuitive to achieve the measurement for displacement field and strain fields.