Abstract: This paper introduces image-matching principle of phase correlation algorithm. A new method based on deletion of pseudo particles and fast computation of correlation after hierarchical matching is presented to solve the problem of false matching and fast sub-pixel computation. A sub pixel precision matching set is obtained by using improved matching flow, which needs 2D-FFT only once. Finally, the improved Phase-based processing has been tested on synthetic particle images, and experimental result demonstrates the effectiveness of the optimized matching method.

Key words: Image Matching, Phase Correlation, FFT, PIV, Fast Computation, Optimization

摘要： 介绍了相位相关法的匹配原理。针对在粒子测速系统(PIV)应用中由于粒子运动而引起的不匹配以及亚像素快速求解等问题，提出了分层匹配后“伪粒子”的去除方法和相关运算的快速求解算法。使用优化的匹配流程仅需一次2D-FFT便可准确求出两幅粒子图像的亚像素级匹配点。最后，使用合成的粒子图像进行了算法验证和效率分析并给出了仿真结果。经验证，优化后的相位相关匹配具有高效性与可靠性。

关键词: 图像匹配, 相位相关, 快速傅立叶变换, 粒子图像测速, 快速计算, 优化