Research And Application Of Data Matrix Two-Dimensional Code Recognition Algorithm

With the extensive application of Data Matrix code in industrial automation field,automation production has put forward higher requirements for the recognition accuracy and recognition efficiency of Data Matrix code.Therefore,more and more research on Data Matrix code location and recognition algorithm has been the focus of research scholars and technical engineers.In this thesis,the related algorithms of Data Matrix code recognition are studied and improved.The recognition algorithm mainly includes image preprocessing,DM region initial positioning,precise positioning and decoding error correction.In the thesis,lots of algorithms involved in DM code recognition are studied and the advantages and disadvantages of each module are analyzed,according to the code image characteristics and giving full consideration to a variety of complex situations.The algorithm is improved by combining with the advantages of different algorithms.The improved algorithm is applied to the Data Matrix code recognition.Firstly,the DM image is preprocessed,followed by gray scale,filtering and binarization.We compared the three binarization methods of bimodal method,iterative method and maximum interclass variance method(OTSU)according to the characteristics of DM image.Then,in the initial positioning module of DM code,the localization algorithm based on regional significance detection is studied,and the initial position of DM code is realized by HC contrast and RC significance respectively.In contrast,we have achieved the initial positioning of the DM code by using a more efficient algorithm based on mathematical morphology and selecting the right area according to the DM area geometric characteristics.In the following precise positioning module,the edge tracking algorithm is improved,and the "L" characteristic line is quickly and accurately determined by the cross search method.The Hough transform algorithm is used to detect the boundary of the DM code linearly and the geometric transformation and the perspective transformation algorithm are used for rotation correction.At last,the precise positioning area is divided and sampled to obtain codewords,using the Reed Solomon error correction algorithm for codeword error correction,and then the DM code information is decoded to data of products.In the thesis,the module algorithm is implemented and the system user interface is constructed in the DM code recognition system,the identification process and the decoding result are also displayed in the recognition system,and we have carried out the test analysis of the recognition effect in the last.The experimental results show that the recognition system can meet the requirement of rapid and accurate identification of DM code in the actual complex situation,and the recognition rate is over 95%,the recognition speed is below 200 ms,which has a good practical application value.