Research Of Flexible Automated Production Line Wheels Model Automatic Identification System

The computer vision recognition and detection technology has begun to show its value intoday’s manufacturing autom-ation, intelligent tide of reform. This is mainly reflected that inthe past, some work such as distinguish kinds of product and control items put spacing couldonly rely on the human eye to finish.But now can rely on the computer and industrial cameracomposition "bionic vision system"-machine vision recognition system to complete. Machinevision recognition system not only has the computer high intelligence and the advantages offast response but also with the human eye fast target capture and imaging capability.Machine vision recognition system is composed of CCD camera and a high performancecomputer and image processing program code development software component. The CCDcamera is mainly used for acquisition target image and its conversion into digital signal.Computer and software is responsible for acquisition of digital image signal processing, thusobtains the target image recognition of the characteristic parameters.In This topic Matlab software is the development platform for identification system.Basedon the reference to the domestic and international wheel recognition technology information,all aspects of the model to identify the wheel as an object, respectively, carried out the researchand experimental demonstration, the final completion of the automatic identification system ofthe set of wheel.In this subject, the wheel identification system is divided into four primary identificationunits and the identification of five sub-modules,Those four identification units are:1.the imageinformation obtaining,2. image preprocessing,3. the image feature extraction,4. imageclassification and identification. In accordance with the order of the recognition of the wheel.First, study non-linear camera calibration method and camera calibration method according toZhang Zhengyou, making calibration target camera calibration process. Second, for all kindsof noise present in the image and background interference, image de-noising algorithm, graysegmentation algorithm were compared, the image is converted into binary image featureextraction can be used. Third, the study feature extraction outer contour size, the same moment,the surface of the number of holes feature extraction, to ensure that these algorithms to extracteigenvalues meet the characteristics of the four principles and reliability. Last, thecharacteristics of wheels feature extraction to, written in Matlab platform that can meet theactual needs of the recognition classifier.Each recognition unit of the system is integrated in the GUI interface, the wheel automaticidentification system has been proved feasible and reliable selection of the various algorithmshas good stability by experiments.