Estimation of microstructure movement underflow using motion vectors in video

Abstract

In many industrial and biological applications, chemical and biological fluid viscosity is a crucial material property that needs correct measurement. Viscosity testing is essential for decision-making, process improvement, and ensuring the quality of the final product. Viscosity measurement has been done in various ways, offering alternatives for various fluid kinds and application needs. These techniques include improvements in microfluidic technology, rotational viscometers, and flow-based technologies. For improving performance, research, and development across numerous sectors and biological domains, precise viscosity assessment is crucial. In these methods, one approach is to use a micropillar-based microfluidic viscometer. The micropillars on this device bend as the sample fluid is injected through the inlets. The quantification of the displacement of micropillars is utilized to determine the fluid's viscosity. For this purpose, the displacement of micropillars for glycerol/water solutions with known viscosities ranging from 2 to 100 cP is recorded using a camera. Then, the displacement results are used to produce calibration curves. The determination of the viscosity of the sample fluid is ultimately achieved through the mapping of observed displacement during the experiment onto these calibration curves. In captured experiment videos with the sample fluid, the displacement of pillars is measured for this purpose using ImageJ, an image processing program. Results obtained using this method are precise. The disadvantage is that using ImageJ to calculate displacement takes time and requires manual work. In this thesis, we propose a novel motion vector-based micropillar displacement measurement method. In this method, the motion vectors are extracted from the video recorded during the experiment. The motion vectors provide the micropillar displacement information without a need for complex computation. We use FFmpeg software to extract the motion vectors to a text file. Then, we use Python to find the displacement of pillars automatically. The displacement results are compared against measurements obtained by manually analyzing the captured images using ImageJ software. ImageJ measurement results are used as a reference to calculate the accuracy of our proposed method. In our experiments, we used 9 videos with different flow rates ranging from 60 ml/hr to 135 ml/hr and different viscosities ranging from 25 cP to 75 cP. We used ImageJ data as a reference to determine the method's accuracy when determining the accuracy of the suggested methods for 9 videos. Regarding ImageJ, our proposed motion vector-based method provided an average accuracy of 90%. We saved a lot of time by using these techniques because we didn't need human assistance. With this method, we were able to process data about 32.72 times quicker than with ImageJ.