Research On Irregular Character Detection And Recognition In Product Management

Product packaging or workpieces are printed with label information such as origin,production date,anti-counterfeiting identification,and safety standards.The label information is related to the product itself.Analyzing and tracing product label information can determine the quality and safety of the product,facilitating the improvement of product management.With the continuous development of technology,product management is closely related to information technology,and intelligent detection and recognition of label information would be a top priority in achieving product information management.In the production process of products,irregular shaped workpieces such as curved surfaces or rings are always produced,and the non-standard use of products can cause distortion and deformation of their labels.These situations can make the label information difficult to detect and recognize.Regular characters are generally easy to detect and recognize,but irregular character detection and recognition are more difficult.Therefore,this article conducts research and analysis on the detection and recognition of irregular characters,facing a series of problems: Firstly,the characters on the workpiece are not clear,and some character details will be lost,which can increase the difficulty of subsequent character detection and recognition.Secondly,If the characters on the object are distorted and deformed,it will lead to defects in the inkjet label.Therefore,it is difficult to effectively detect them using traditional character detection techniques.Thirdly,If the characters are in an irregular shape like a ring,it is difficult to use traditional methods to detect and recognize label information effectively because the characters are arranged unevenly and horizontally.Based on this,this article conducts research on irregular character detection and recognition algorithms through deep learning methods and obtains the following conclusions:(1)An improved multi exposure fusion algorithm is proposed for images with unclear details lost.This method can effectively perform image preprocessing,not only overcoming the problem of uneven exposure,but also enhancing the detailed features of characters in the image,preventing the loss of image feature details and improving the clarity and contrast of the image.(2)A PSO-TSA fusion ICMN model is proposed for characters that are prone to distortion and deformation.This method can effectively detect bent characters.Based on the final experimental results,this method can greatly improve the accuracy of detection and have significant advantages in terms of H-means performance and efficiency.(3)The boundary point generation strategy and image correction method are proposed for the research of circular character recognition.This method can effectively process the characters in circular shape,so that the recognition accuracy of 52.2% can be improved to 96.8%,and the running time can be reduced from 813.84 ms to 407.29 ms,greatly improving the accuracy and efficiency of character detection and recognition.Figure [39] Table [16] Reference [83]...