High-precision Dynamic Contact Angle Measurement Based On Gaussian Mixture Model

Dynamic contact angle refers to a series of contact angle values that change transiently with time when the droplet is tilted or the volume is changed.It is an important characterization method for cutting-edge research in materials science,and its measurement requirements are increasing.Therefore,improving the accuracy and efficiency of dynamic contact angle measurement through machine vision technology has an important application value.Aiming at the problem that the binarization and edge detection algorithms in traditional static contact angle measurement are not effective in identifying dynamic contact angles,this paper studied the use of Gaussian mixture model background modeling and improved algorithms to extract droplet dynamical edges,and established one systemic dynamic contact angle measurement platform.The main research conclusions are as follows.First,this paper analyzed the influence of the main sub-links and important parameters of the traditional Gaussian mixture model on the detection of moving targets,and designed a Gaussian mixture model algorithm with adaptive adjustment of learning rate.By automatically judging the droplet spreading stage of the video sequence,taking different learning rates α for different stages,and updating the number of Gaussian components according to the previous and subsequent multi-frame judgments,the algorithm showed suitable for slow moving targets.By comparison with mainstream moving target detection algorithms,the experimental results showed that the algorithm in this paper can better detect the edge image of droplets when processing the spread video of droplets,and has a higher recognition speed.Secondly,the preliminary droplet edge image obtained is improved,and random edge points are used to fit the droplet edge image using the least squares circle fitting algorithm to obtain a high-quality droplet edge contour image,while Canny edge detection algorithm is used to obtain droplet edge images for static droplet images.The hypsometry measurement method and circle fitting algorithm were used to calculate the contact angle values of the droplet edge images in the above analysis.Third,the software and hardware platform of the dynamic contact angle measurement system based on the improved Gaussian mixture model was constructed.The Development was based on the C++ language and Open CV vision library under the Visual Studio integrated development environment of Windows platform.Real-time acquisition imagings of liquid droplets with a diameter of1-2mm were obtained through optical platforms,high-speed cameras and other hardware.And the measurement system also has functions such as moving target detection,frame-by-frame image,boundary detection,machine measurement,etc.The software interface design combined with MFC is convenient for updating and optimizing the image processing algorithms involved,reducing the dependence on hardware level,and still maintaining a high accuracy rate for videos with poor shooting effects.Finally,the data processing and analysis of measurement accuracy were performed.Aiming at the problem that light or shadow may distort the measurement results in some frames in the droplet spread video sequence,a boundary mutation judgment process was added.For the frames determined to be affected by light or shadow,the standard angle method was used to calculate the contact angle value instead.Ten selected droplet spreading video sequences are used for verification,and feature point detection contact angle measurement algorithms are used for comparison.The algorithm proposed in this paper is superior to the feature point detection algorithm in indicators such as root mean square error,average absolute error,and standard deviation.