The Research On The Intelligent Inspection System For Foreign Subjects In Transfusion Bottles

The on-line inspection system for foreign objects in a bottle of transfusion solution is presented in this thesis. Based on the background of inspection foreign subjects in transfusion bottles all over the world, the system structure was designed and described. The system structure, the inspection algorithm and the software implementation were mainly researched.Firstly, the application of machine vision in industry inspection is represented, and the type of machine vision system is analyzed, whose characteristics are compared. Then, the illuminating system, cameras and lens that are concerned with the inspection of foreign objects in transfusion bottles is introduced.Secondly, this thesis analyzes the difficulties in the design of intelligent inspection system for foreign objects in transfusion bottles. The main structure of this system is presented at last. The whole system is composed by three rings. The first ring is an in-feed star wheel that carries the transfusion bottles to the living ring, the second ring, where foreign objects are inspected. During the inspection process, at first a bottle filled with transfusion solution moves along a live ring while rotating around its own axis rapidly at a high speed. A vortex rises in the bottle that drives foreign matter from the bottom of the bottle to circle to the surface in the solution. Later, the rotation speed of the bottle becomes slower, and the solution vortex gradually disappears. When passing in front of a camera, the bottle stops rotation abruptly, and the live ring stops running. At this time, only the solution still moves; inside foreign matters because of gravity and inertia move down to the bottom of the bottle, and bubbles rise to the surface of the solution. After stopping for a short time, the transfusion bottle moves on, and then enters into the sorting unit, which is consist of the third ring.As for the inspection algorithm, camera calibration, motion inspection and the identity of foreign objects are discussed. A normalized cross correlation is used to obtain the frame, based on that the background is estimated from the first frame by inter-frame difference, and small moving objects are extracted from each frame. Next after grey dilation and thresholding, an adapted mean shift tracker is adopted to find falling foreign objects.At last, based on the analysis of software used in machine vision, the software for inspection of foreign objects in transfusion bottles was developed, and the main design and structure is presented. The implemented software is given in this thesis.