A Study On The Method Of Checking For Label Of Medicine Bottle

Currently, the preparation process of intravenous medicine is done by dispensary staffmanually. In preparation process, the dispensary staff need to check the name of medicine,the dose, the time and other information repeatedly, which lead to large workload and proneto error. To solve these problems, this paper studied a method to realize the automaticchecking for the label of medicine bottle.This paper introduced the literature at home and abroad related to the checking methodfor labels of medicine bottle, described the image processing method for the recognition oflabel image and known the present development situation of image pattern recognition athome and abroad. The study object in this paper is the curved label image, on both sides ofthe label image will appear distortion of character compression and existing imageprocessing method can not accurately recognize this type of label image. Therefore, thispaper study a method based on Matlab image processing which can be used to therecognition of curved label image. This method mainly consists of two parts, one is toestablish a standard bottle label image template library. Select the clear label of medicinebottle to expand and convert it to a binary image after preprocessing. Correct the unfoldedpanel label image using a certain algorithm and covert it to the curved label image. By thismethod, process the label image for checking one by one and use the label image aftercorrection as the template image in order to establish a standard bottle label image templatelibrary. The other one is to match the label image acquired real time. Process the targetimage acquired real-time use the same image preprocessing method, then select the unit areaand calculate the statistical characteristics of pixel value in each unit area. Calculate thefeature matrix of target image and each of the template images respectively and calculate thesimilarity of feature matrix between the target and template image. The informationcorresponding to the template image which has max similarity with the target image is thematching results. This paper designed a hardware system, including the image sensor, theimage sensor interface circuit, the display circuit, the serial port circuit and the core controlcircuit. The type of camera in this circuit is OV9655and the core control device is theembedded single chip microcomputer of STM32series. This system use serial port totransmission the image data acquired to the upper computer, use the Matlab to compile the program in upper computer and use Matlab GUI to design the user interface.This paper used multiple sets of label image of medicine for the experiment anddiscussed the effects of result in different experiment condition. Improve the algorithm basedon the experiment result and the experiment results show that this algorithm can accuratelyachieve the automated checking for label of medicine bottle.