Barcode Localization Algorithm In Complex Background Based On SVM

With the rapid development of networking technology,barcode technology has been used more and more widely,it not only involves in transport and management,but also takes part in our daily life,such as phone payment and data download.In conclusion,barcode technology is more closely related to our lives,in order to further tap the potential of barcode and create more convenience for our economic life,we should find the method to quickly detect the barcode in complex background.According to the existing barcode localization algorithm,I find some problems:Firstly,It is difficult to identify the barcode in complex background,such as tables and text,their texture is easily confused with barcode;Secondly,Because the different kind of barcode has different encoding formats and texture,so it is difficult to identify the barcode in the picture at the same time.In this paper,I propose barcode localization algorithm based on support vector machine,the process is as follows:(1)Firstly,make the bacode image sample,in the collection of positive sample,guarantee the proportion of barcode is as high as possible;(2)Extract the Local Binary Patterns(LBP)feature of the barcode sample,then train them and output the classifier;(3)Make use of the sliding window technology and a certain step to get the block in the picture,then get its classification;(4)To gather all the blocks belong to the barcode,then merge all the block connected.The experiment results show that the algorithm proposed in the paper can get better effect on the time complexity and localization accuracy.