Visual Detection Method for Missing Infusion Bag Pipeline

Abstract

As an essential medical device, a bag infusion set is often used for intravenous infusion, and an infusion bag is an essential part of the bag infusion set. Due to the unavoidable defects in the production process, quality detection of infusion bags is critical to ensure the use quality of the infusion set. In this paper, we adopt a machine vision system to inspect the assembly quality of the lanyard and dosing interface of liquid bag assembly and conduct in-depth discussion and research from image acquisition, a defect detection strategy, and a defect detection algorithm of a vision system for two defects of lanyard missing and dosing interface missing. The design of the image acquisition auxiliary mechanism is realized to solve the complex problem of image acquisition due to the irregular shape of the liquid bag assembly; based on determining the defect detection strategy, the algorithm study of contour extraction is finally completed through comparison experiments to extract a precise contour of the liquid bag piping area; finally, the virtual straight line method is proposed and combined with the ROI selected according to the position feature of the outer rectangle of the contour in this paper, the count of the number of contours is completed, and the defect detection goal is finally achieved. The pipeline defect detection rate of the method proposed in this paper reaches 100%, which can perfectly replace the existing manual visual inspection and reduce the employment cost of enterprises.

Improved Faster R-CNN Network for Liquid Bag Foreign Body Detection

“…Vignetting is generally classified into mechanical vignetting, optical vignetting, natural vignetting, and pixel vignetting, and vignetting correction methods are generally based on the "flat field correction" theory. The theory considered that the real image 𝐼 with a vignetting effect was composed of an image V containing the attenuation factor of each pixel and an ideal image 𝐼 with a uniform color plane, as expressed by Equation (1). This vignetting phenomenon was more prominent in vision systems with large fields of view, and the liquid bag foreign body detection image in this paper suffered from severe vignetting, as shown in Figure 3.…”

“…This can result in misdetection and the leakage of fluid bags. To address the issue of missing pipelines, researchers have combined machine-vision technology to detect the absence of pipelines in fluid bags [1,2].…”