Liquid Level Detection of Syringe Pump and Syringe Based on Machine Vision

Abstract

When a nurse uses a syringe pump to infuse the patient, the nurse needs to contact the patient several times at a close range. This article proposes a method to detect the infusion of the syringe pump to reduce the frequency of contact between the nurse and the patient and reduce the labour intensity of the nurse. Using machine vision technology, filter and Prewitt operator are combined to process the acquired syringe pump pictures to obtain the accurate liquid level position. The image is captured by an industrial camera. After grayscale processing and image filtering, the Prewitt edge detection algorithm is used to obtain a clear edge image, and then the corner detection algorithm and pixel coordinate display are used to detect the liquid level of the syringe without external auxiliary measuring equipment. After performing multiple tests on 26 different liquid level images of a 50ml syringe, the results show that proposed method is effective to measure the liquid level in a non-contact way. The test result shows that the relative error is When the scale is 2ml to 16ml, the relative error range is -66% to -0.63%. When the scale is 18ml to 50ml, the relative error range is -1.86% to -1.19%. Among the 26 pictures, the absolute error of 22 pictures is within ±1ml, and the absolute error of 19 pictures is within ±0.5ml.