Research And Application Of Embedded Two Dimensional Bar Code Recognition Technology

At present, the three standard international trade methods are: bar code, Electronic DataInterchange, container. Bar code technology has been widely used in transportation, retail,manufacturing industry, tele-communication systems and so on because of such advantages as itshigh input speed, accuracy, reliability and stability, strong practicality, low cost and easyaccessibility. In addition, bar code has been employed in the library management, asset managementand tracking, bank notes, company file management and official document management ofgovernment authorities. So far, China is still in the phase of developing country, it is of greatsignificance to develop such economical recognition equipments which greatly facilitate thedomestic dimensional barcode technology and industry development.In order to meet the industrial assembly requirement of barcode recognition, the authors firstdesigns the relatively general two-dimensional barcode recognition system including externalhardware and software architecture. Then the author makes explanations on each section withemphases on the research on the recognition software and establishing the overall consciousness forrecognition.In order to carry out the software research, this article focuses on the specific two-dimensionalbarcode—the study on the reading algorithm with Data Matrix. In as for the preprocessing oftwo-dimensional barcode’s image, the article makes a contrast among such common methods asgray, filtering de-noising and two value binary. According to the comparison results, the mostsuitable way to preprocess the Data Matrix’s image is median filter of the3*3template and theOSTU binary.In order to strengthen the versatility with algorithm, this paper designs two recognitionmethods according to the different backgrounds of Data Matrix. As for the simple background ofData Matrix, after the preprocessing including the gray, gray stretch and3*3median filtering, wecan make edge detection through Roberts operator, after getting the image edge, we can makeregional mark with the eight adjacent domain recursive method. Furthermore, using the outline ofData Matrix for rectangular characteristics, combining the regional area, shape parameters, we canextract the rough region of barcode through designed unique barcode thick localization algorithm.According to the positioning graphics’ L shape characteristics of Data Matrix barcode, we cantransform the two-dimensional border into a curve with its distance as angle and then define theangle and regional coordinates of barcode. This boundary marker is the first application into theposition of barcode, which has certain academic value.At the same time, we also design the two-dimensional barcode recognition algorithm concerning the complicated background. After the gray transformation and two-value binary, wecan not only remove partial noise of the image but also rule out many barcode interference regionsthrough applying the3*3template into the mathematical morphology processing of images. As forthe coarse positioning of barcode, we adopt the position method applied in simple background toextract the regions of barcode. However, considering that there are more disturbances in complexbackground, we employ the fine position method using Hough to transform the longest, two verticalstraight lines detection to determine the angle and position of barcode, which can achieve thepurpose to remove the disturbance regions.In order to improve the accuracy of the data collection, this paper also employ the gray curvebased on barcode to determine the positioning graphics. After the acquisition of barcode’s generallocation, we use the least the score line and column as a goal line and list after reverse sent betweenthe first four gray-scales respectively with four columns, and then draw gray curve regarding thegoal line and all of the data. Peak curve corresponding black data module center, peak and valleycorresponding white data module center, we can obtain the positions of peak and peak valleythrough a grayscale curve. At last, we can extract the data of two-value image according to thebinary image grid methods of the positions. This method is accurate and fast with low error rate,which is very helpful in the speed of the algorithm.The current subject adopts an engineering idea: first determining the overall scheme and thenusing Matlab simulation platform to carry out the simulations. Regarding the barcode recognitiontest of barcode images with the pixel as640*480, we obtain that the positioning time is0.26s andthe positioning rate is98.9%in the simple background. Through the experiments, we effectivelyverify the accuracy, practicability and reliability of the original algorithm. Meanwhile, thepositioning rate of bar code in the complicated background is more than98%and the recognitionrate is above98.5%, which can satisfy the industrial assembly requirements of two-dimensionalbarcode recognition.