Laser Diode Arrays Coding Recognition

With the development of image processing and pattern recognition，computervision is widely used in various fields of production and people’s life. Laser codedrecognition as one of optical Character recognition technology，mainly used for laserarray testing system，as a part of laser array test system help with other part completeall the testing process.This paper use the laser array from Genuine as the main object to study，firstanalysis characteristics of the laser images and critical problem mainly needed tosolve. Do research and deployment deep in some of these technologies. Propose apackage of solutions for laser coding recognition. Compare various parts of thecontent and algorithms in the program，so the research is valuable in either theoreticalor application. Depending on the research，summarizes this paper as follows：1. Research viable laser image preprocessing methods，find the algorithms toprocess the picture affected by noise or with low contrast to make it more easy toidentify.2. Proposed a method using edge density characteristics for the laser imagesegmentation. Proposed two measures to build connected components form the edgeof the dot density. With the help of these components segmentation is completed. Thismethod has a good result in processing the image who’s forward color andbackground color are too similar，or who’s background has large blob of noise.3. According to the position relative to the laser array of the code. Proposed aregional segmentation method based on projection. Calculate the feature of thesegmentation finish text detected。4. For coding pictures with illumination uneven, serious noise, A textsegmentation method is been proposed variance using gradient projection, Adjust thecharacter with the inherent characteristics of the divided region to be the bestsegmentation.5. Design a characters classifier based on support vector machine. Analysis andcompare different feature selection brings impact on the recognition rate.Study theparameters of classification model，carry out experiments on different parameterstrends impact recognition rate.