基于块匹配的粒子图像测速算法
Particle Image Velocimetry Algorithm Based on Block Matching
-
摘要: 在粒子图像测速(PIV)算法研究现状的基础上,提出基于块匹配的PIV算法.针对算法中的匹配区划分和匹配准则选择,给出了匹配区划分的一般准则,即匹配区尺寸必须显著大于粒子直径;同时提出了以最小绝对误差/最小均方差匹配准则替代传统的互相关准则的PIV算法.实验证明,采用最小绝对误差和最小均方差准则替换传统的互相关评价准则,在保证估计精度基本不变的前提下,算法的运行时间减少超过50%,运行效率提高一倍以上;实验结果也证明了匹配区划分准则的正确性.Abstract: Reviews the research status of the Particle image Velocimetry algorithm based on a new block-based matching algorithm of PIV have been proposed.For two key algorithm-to match the zoning and match the criteria for selecting article gives general guidelines to match the zoning,the size of the matching area must be significantly larger than the particle diameter;MAD / MSE matching criteria for alternative traditional inter-related criteria for PIV algorithm.The experiment proved that the use of the MAD and MSE criteria to replace the traditional cross-correlation evaluation criteria to guarantee the estimation accuracy is essentially the same premise,the running time is reduced by more than 50%,operating efficiency more than double;experimental results also clearly proved the correctness of the match the zoning guidelines.