Abstract: To overcome the drawbacks that solutions based on traditional Digital Image Correlation（DIC） methods are possible local optimal values, the Particle Swarm Optimization（PSO） algorithm based on swarm intelligence is introduced into the DIC method. Displacements of points in a speckle image that is displaced an amount are calculated, and the DIC method based on the PSO is verified. If a few subsets in translated or deformed image are similar to a subset in initial image, the correlation coefficient distribution can possess some peak values. Displacements of two points in this kind of images mentioned above are calculated, and the trajectories of particles’ motion are demonstrated. It is found that the DIC method based on the PSO algorithm has strong global searching capability in the initial stage of computation, and the method has local searching capability in later stage of computation. Results show that the method can escape from being trapped into local optimal value. Effects of the subset size, the number of particles, the maximum velocity and the times of maximum of iteration on the calculation time are investigated.

摘要： 为了克服基于传统数字图像相关（DIC）方法的结果易于陷入局部最优等缺点，将基于群体智能的粒子群优化算法引入到DIC方法中（未考虑亚像素插值），对一幅散斑图平移后的位移进行了计算，验证了该方法的正确性。在搜索域内，当运动或变形后散斑图中若干目标子区与样本子区比较相似时，相关函数可能有多个极值。对这种情况下的点的位移进行了计算，观察了相关搜索时粒子运动的轨迹。研究发现，在迭代初期，该算法具有较强的全局搜索能力；在迭代后期具有较好的局部搜索能力。计算结果表明，该算法可以跳出局部最优；研究了样本子区尺寸、粒子数、粒子飞行的最大速度和最大迭代次数对计算时间的影响。