Realization Of Code Identification Based On The Character Display Principle Of Electronic Instrument

With the rapid development of science and technology, computer visionis applied in many fields and especiallly in character recognition.For example,the automatic identification system for licence plate has come into practicalapplication. Based on the requirements from Ministry of Railways ofPRC for the process from steel-making to warehousing onheavy railproduction line, the lab and Wuhan Steel Plant have made joint efforts todevelop an automatic identification system for on-line steelbillets on heavyrail production line independently to meet the urgent demand of automatictracking. The application of the system can not only effectively reduce thelabor intensity, but also improve the overall automatic productionlevel. Therefore, character recognition technology plays a significant role inthe development of automated management and manufacturingintellectualization.The thesis sets the automatic identification system for on-line billetsonheavy rail production line in Wuhan Steel Plant as the background, andstudies on the improvement of the original template matching algorithm witha purpose of enhancing identification efficiency and automation level. It putsforward a method of character structural feature extraction and recognitionwith the help of electronic display identification. Thismethod transforms thecharacters after structural feature extraction to standardize the charactersfollowed by the realization of code identification based on the characterdisplay principle of electronic instrument. It overcomes the defects of currenttemplate matching recognition algorithm, moreover, it owns advantages likehigh recognition speed, and better recognition rate by enablingto recognize separated and defect characters.The entire process for billet character recoginiton system includestwo major parts: pre-processing and character recognition, among which pre-processing involves four steps of denoising, positioning, segmentation,thining while character recognition relates to three aspects of projectiontransformation, electronic display and code identification.The method distinguishes from template matching inperformingprojection transformation on characters after thiningpre-processingstarting from the structural features of characters and thentransforming into the standard characters similar to digital display accordingto the results of projection transformation. Finally the billet characteridentification is achieved by comparing the electronic character encodings.The experimental result shows the method of character structural featureextraction and recognition using electronic display proposed in this thesisowns advantages like fast identification speed, strong stability and highidentification rate, which can provide a more efficient identification methodof real-time detection for on-line steel billets.