The Research And Reality Of Character Recognition In Engineering Drawing

In recent years, with the continuous progress of the computer technology, the application in the related field is becoming more and more popular. Character recognition technology is one of this applications. With continual research efforts,character and digital identification technology is more and more causing the attention of people. At the same time, the demand of the character recognition technology is increasingly rising, which also gives the significance of the research about the character recognition.Character recognition technology has been very mature, many practice has been used in some fields. But the application in the field of engineering drawings is blank.There is still some lack of the related technology. The content in this article is a summary of the research and practice about character recognition technology based on engineering drawings. I wish it could be helpful to the related people.This program of the article comes from Jilin university.The program means to use computer automatic recognition of characters in engineering drawings instead of artificial recognition. In reality, the number of the drawings is huge and the content is multifarious. Manual input often accompany a lot of work, high cost, tediousness and error-prone. So replacing the artificial recognition with the intelligent identification has a significant meaning. This article provide the improved algorithm of the feature extraction of numbers and letters. Because the shape of characters and the numbers is similar, this paper introduced a improved algorithm to this issue, also to solve the problem of the fault recognition of the similar characters. There are many other problems, not in this list. The main of this article includes the following several aspects.First of all we should do some pretreatment,which mainly includes the following several aspects.Graying the drawings, which turns the three-channel images into a single channel image, used to reduce the original image data, facilitating to the subsequent processing of the images.Image binarization, which can make the extraction of the information in the image convenient, binary image in the computer identification can increase the recognition efficiently, increase the recognition rate, eliminate noise and at the same time decrease the rate of fault recognition.After the pretreatment, we can start on related characters recognition. Character recognition method is consisted of pattern recognition, feature extraction and neural network.This article mainly involves pattern recognition and feature extraction.Character recognition of engineering drawings has its own characteristics.The number of engineering drawings is huge. because the drawings can be done in different time, from different hands, a criteria, using which creates a method for identifying the part drawings may not apply to another part of the drawings.The content of every engineering drawings is relatively complex, also with much noise, how to effectively remove the noise to improve recognition rate, decrease the rate of fault recognition is a important problem.Because the huge number of the drawings, the original data becomes big.So the recognition of characters in the engineering drawings also need to reduce the time complexity of the algorithm.This article will summarize the problems in the project and the solution one by one. The methods and results have been checked in actual project.