Machine vision-based automatic liquid dispensing metering technology and system for microfluids

In modern scientific research and industrial applications, the rapid, automated, and accurate measurement of micro-liquid volumes added to reaction or detection containers is a critical need. Traditional methods for measuring micro-liquid volumes often suffer from insufficient accuracy, low stability, and are prone to interference from bubbles between microliquids and residual droplets in the transmission pipelines. To address these issues, this paper proposes an automated microliquid metering method and system based on machine vision. The system comprises optical imaging units, drive control units, image processing units, metering algorithm units, and calibration units. By optimizing the optical imaging setup, the brightness and contrast of the liquid in the metering field are enhanced, ensuring the accuracy of the volume measurement. Additionally, image processing algorithms are employed to segment the liquid section, and its length in the pixel coordinate system is extracted as a representation of the volume, effectively eliminating the interference from bubbles in the image. Finally, calibration-based measurement methods and direct measurement methods based on homography matrix scale transformation of marker points achieve metering accuracies of 98.2% and 98.3%, respectively. Compared to traditional industrial micro-liquid metering methods, this approach effectively overcomes the impact of bubbles on measurement accuracy while offering greater stability and reliability.