Development of Real Time Measuring Technique for Particulate Objects in Multi-Phase Flows Utilizing Image Analysis

该项目旨在测量在拥挤条件下移动的颗粒物体（如颗粒和微小气泡）的数量和速度。为此，粒子跟踪测速技术是最合适的，因为该方法可以测量单个粒子的速度。但是，该方法的现有算法在粒子密集分布时通常较弱。本研究主要是在二值图像相关方法的基础上提出一种改进的PTV方法，该方法对分布较密的粒子有效。该方法的基础是追求粒子簇之间的相似性，计算一个特殊的相关系数，该系数针对二进制数据进行了优化。二值图像相关方法可以在很短的时间内找到两个连续图像之间的粒子间对应关系，然后通过分析连续图像来跟踪每个粒子的运动轨迹。当不断追求运动粒子的轨迹时，这些粒子可以被识别出来。即使一些粒子在某些图像中被观察到是一个粒子，它们也可能在不同的时刻和图像中被分离。根据上述概念，开发了一套分析软件，并证实了在粒子密集分布的情况下，新系统的分析效果远远优于原系统。但是，它仍然存在一些改进的问题，如更快的分析，减少错误的测量和积累更多的实验示例以获得更强的性能。在未来的研究中，我们将对图像进行灰度分析，以区分组成大簇的单个粒子。

This project aims to measure number and velocities of particulate objects such as a particles and tiny bubbles which are moving in crowded condition. For the purpose, the techniques of particle tracking velocimetry is most adequate, since the method can measure velocities of individual particles. But, existing algorithms of the method are commonly weak in the dense distribution of particles. In the present study, main effort is paid to develop a improved PTV method from the binary image correlation method which is effective for densely distributed particles.The method stands on pursuing a resemblance between clusters of particles calculating a special correlation coefficient which are optimized for binary data. The binary image correlation method can find particle to particle correspondences between two successive pictures in a very short time, then each particle trajectory can traced by analyzing sequential pictures.While continuously pursuing trajectories of moving particles, those particle can be identified. And even if some particles are obesrved as one particle in some pictures, they could be separated in different moment and pictures.Along with the above mentioned concept, an analyzing software is developed, and it is confirmed that the new analyzing system works far better than the former system for a condition of densely distributed particles. But, it still remains some issues of improvement such as faster analysis, reducing erroneous measurements and accumulating more experimental examples for stronger performance. In the future study, we are going to apply gray level analysis of the pictures to distinguish individual particle consisting large clusters.