数字粒子图像测速技术中目标分割算法的实现

数字粒子图像测速(DPIV digital particle image velocimetry)关键技术在于提取流场中粒子的运动信息，涉及连续两幅图像中粒子群的对应(互相关性)等图像处理问题．本研究提出改进的互相关算法，将数字成像系统的连续两幅图像中的目标搜寻区域进行自动分割，将其划分为互不交叠，各自具有一致属性的区域图，然后依次将第一幅的判读小区在第二幅的大搜索小区中移动，搜索与第一幅判读小区最匹配的粒子图像，此方法完全适用粒子速度在1．2～3．5m/s范围的粒子流速度测量，运算速度和精度均满足实际需要．

A new object division algorithm for digital particle image velocimetry processing technique

The digital particle image velocimetry (DPIV)technology is mostly based on the velocity vector information in particle stream field from image date, deals with the image processing of particle flux in two seriel maps. An improved auto-correlation algorithm for pairs of two separate images recorded directly with a CCD camera is proposed, which is decomposed into search windows( small sequare regions ). Its flow region is identical for defining a non-correlation window of the operation, then, displacement estimation small objects regions in the first image relative to the second one. The results indicate that the new developed algorithm to allow particles velocity 1.2-3.5 m/s, and the accurate and processing time are more significant increase than the conventional cross-correlation method.