A Real-time Double Emulsion Droplets Detection System using Hough Circle Transform and Color Detection

The object detection based on computer vision algorithms has a wide range of applications in many domains. This paper introduces a real-time detection methodology of 3D printed double emulsion droplets. The first step is to directly segment and extract the color of the inner droplets from the current field of view using thresholding operations of inRange functions of OpenCV in YUV space, so that the droplets color and background color can be obtained separately. Secondly, a binary masking has been employed to locate the centre of the mass of droplets and also distinguish the droplets from surroundings. Next, the Hough circle transform is used to fit a circle to extract the true edge when identifying the contour of the external droplet. The droplet size, generated frequency, and determine whether the droplet is successfully formed can also be obtained using Hough circle transform. A series of experiments have been conducted and indicated that our proposed detection system works well on different double emulsion microfluidic devices with different background and droplet colors.