Study Of Inspection System For Visible Impurity In Ampoule Based On Machine Vision

Ampouled medical liquid detection is an important research topic in the field of medicine detection. Nowadays, visible impurity inspection in ampoules is accom-plished by eyes of professional workers in most domestic pharmacy corporations. Considering the inefficiency, the high cost and the unstable precision of eyes inspec-tion, using machine vision to find the foreign bodies in the ampoule is an inevitable trend in the development of ampouled liquid detection techniques. Based on the pre-vious studies, a new multiple image data acquisition and processing system is de-signed and realized, and the principle in visible impurity inspection in ampoules and the algorithm applied in DSP are studied. The main contents of this dissertation are as follows:The new multiple image data acquisition and processing system can read the da-ta from CCD and perofrm the image binarization and the impurities discrimination rapidly, with a TMS320DM642 chip as its hardware core.The ampoule image registration methods are studied. According to the characte-ristics of the ampoule image, a rapid image registration method based on local gray mean square measure is proposed, which is used together with difference operation, image enhancment and binarization to extract the signal of visible impurities in am-poule image.Impurities recognition of the image data signal is also studied. Through search-ing the suspected signal orderly and building the detection window to identify the impurity and its quantity, a signal for sorting signals is given to the mechanical transmission structure to finish the visible impurity inspection and the ampoule sort-ing.This research work can provide a guide in extracting image signals and discrim-ination of key information in data processing, and is of significance for realizing a fast, miniaturized and integrated visibleimpurity inspection system.