The Design And Implementation Of Transcript Entry System Based On OCR

With the high-speed development of society, the education of the people in thisworld to more and more, every day there are a large number of reports informationneed to entry, in order to better improve the work efficiency and reduce labor intensity,effective personnel, we here is an urgent need to solve the problem, is to write a reportcard information quick entry into the computer. A collection of scientific researchsubject, in one part of the report, for example, we have designed a transcripthandwriting recognition using OCR system. Thesis mainly finished the followingwork:(1) Transcripts in line usually have form, because the handwriting is differentfrom person to person, report form is small, so we usually will form after line removalin identification. So this paper used the linear detection based on Hough transformtechnology, combined with the adaptive median filtering blanking technology makesthe straight line.(2) There are a lot of information in the report, such as: student’s name, grade,subject information such as effective information blocks. In order to block theinformation accurately identified, we generally according to the transcript of theimage layout analysis, adopt the connected domain to identify the information block.(3) based on row and segmentation of the characters. Based on principles oftopological structure of Chinese characters, Chinese characters after the sizenormalization, to carry on the elaboration of Chinese characters, such as segmentationmethod effectively solves the Chinese characters in the process of Chinese characterrecognition variation after the segmentation problem.(4) Based on Chinese character refinement method, this paper adopted thedensity, grid characteristics, combining the characteristics of peripheral feature extraction method, based on the extraction of contour segmentation, effective tocharacter feature extraction.(5) After completion of the segmentation process based on the characteristics ofthe characters, according to the density of handwritten character features, grid andperipheral features and the density of handwritten character in the template, grid andperipheral features to extract the characters of the absolute value of the difference, thetemplate matching, complete Chinese characters handwritten character recognition.(6) Report card image, the author of this paper has carried on the processing ofdifferent modules, including image preprocessing, layout analysis and classificationof handwritten characters such as OCR module, the module function betweenrelatively independent, which is based on the modular principle of program design, atthe end of all modules are integrated, and eventually after processing method based onthe database, will we recognize the results and check the inside of the databasecontent, thus finally completed in OCR technology to handwriting recognitiontechnology.