Research On The High_Performance Measuring In Large_Scale Displacement And It's Rectifcation With Video_Image Processing

1.IntroductionDuring the high-performance measuring of large_scale , we usuallyuse the multi-lens from the different angle of view to capture imagein order to enhance the precision of the image. This method dividesthe great size into some small regions and reduces the lens capturingscope. This method cost high, operating complex, the account is big,not meeting the timely requirements.Above all, using the single view measure the large_scale have morepractical significance. This method cost low, operating simply,reducing the time of image processing, meeting the timely requirements.The key technology of the single camera measuring includes distortionrectification, sub-pixel location, video capture.2.The whole design frameworkA typical vision application system includes the light source, theoptical system, the image capture system, the image digitizationmodule, the digital image processing module, the intelligent judgmentdecision-making module and the machinery control module. In the partof hardware designing, we selected the medical piece-viewing lightthat was taken as the light source of the visual system. Japanese Telidigital surface CCD Camera was used as the optical image capture system.We also have a manual 6 time of zoom lens whose focal distance scopeare 8.5mm~51mm, and the view angle of which are 31.3~5.5°. The videocapture card that we selected is the Meteor-II 1,394 black-and-whitevideo card that is made in Canadian Matrox Corporation.In the part of software designing, we want to realize the mainfunction as follows:(1) Real-time capturing image and displaying image: Transmits theimage data that is captured by CCD Camera to the computer through thevideo capture card.(2) Locating and detecting the edge of sub-pixel: Compiles theprocedure with the corresponding algorithm of sub-pixel.Automatically completes the precise sub-pixel location of the goaledge by using software. And it can reflect the corresponding distanceof the goal edge.(3) Tracking and locating the target : In the process of the pullingtest,the software may track the movement of the corresponding goal.And it can also apply sub-pixel to accomplish localization accurately.Accordingly, It can reflect the accurate displacement.(4) Real-time processing data and outputting the record : In theentire experimental process,It can simultaneously display thedisplacement that the target travels and automatically record andprocess .The system software is manufactured mainly by Visual C++6.0Procedure software under the Windows environment. It is based on MILsoftware package that is brought by MeteorII video capture card.In this system design ,we took the toothpick which was fixed onthe stretch test sample as the mark , Selected the backlighting methodto capture the outline of the mark with CCD . Then we could measurethe lateral deformation of the test sample with the correspondingalgorithm. We used the backlighting method to solve the problem becauseit could save the energy and reduce the effect that the externalenvironment influenced to the test result.The overall structure drawing of the system is shown as figure 1.Fig.1 The overall structure of the system3. Camera calibrationIn order to extract the geometry information of the space object fromthe image, we must establish the mutual corresponding relation betweenthe pixel's position and the surface dot of the space object in thephotographic measure. The image module of the photograph systemdecides the mutual corresponding relation. The parameters in the imagemodule of the photograph system are the Camera's parameter. In mostconditions, the parameters must be determined through the experimentand the computation. The process of the experiment and the computationis named camera calibration.As a rule, common image distortion is due to the fact that differentareas of a lens have slightly different focal lengths, object planeand image plane are unparallel and geometry error in imaging system,especially the thick lens used. Especially in the monitor system needdistortion rectification because of the thick lens and wide-angle lensbeing used. Most methods of Traditional distortion rectification arethrough polynomial transform. In order to describe radial distortion,two variables raised to three or more powers must be used. This kindof operation waste too much time, so its application in real-timesupervision is difficult. Considering the real time image disposingfunction of videoextensometer, we advanced one piecewise method basedon the image distortion from the geometry error in the video systemconsisted of CCD camera, the video capture card and monitor. The methodthat divide image district into different areas takes certain spotsin the every region to make one power multinomial fitting. Through thismethod, we reduce the account, saving the calculating time with thesame effect compacted high power multinomial.4. The sub-pixel location of image targetThe principle of the image measurement system is through processingthe image edge to obtain the object's geometry parameter. Obviouslyin the image measurement system, the image edge detection is thefoundation and key. Along with the measuring accuracy requestenhancement, the pixel detection already cannot satisfied the need ofmeasure precision, therefore the new measure named sub-pixel edgedetection was produced to satisfy the need of high accuracy.In the system, edges are detected by gray-level matrix. The basicprinciple of gray-level matrix edge detection is the supposition thatthe actual edge distribution in the true image is consistent with thegray-level matrix in the ideal model that its edges are stepping,namely matrix is invariable. We determine the actual edge positionthrough this relation. In the processing of the image precise measure,we can advance the one dimensional ideal stepping edge, then accordingto the characteristic that the gray-level matrix to maintain theinvariable principle, carries on the sub-pixel localization5. Testing data and error analysisAccording to the stability, the distortion rectification and thesupervision for constant speed motion,the metal and nonmetal extendingtesting data analysis, there are several factors to influenceexperiment output result, such as the number of points searching edge,data smoothing number, capturing frame number, and extending speed.The source of error produced in the digital image measurementsystem have very many. Only search these source of error and carry oncorresponding processing and the analysis, we can acquire real testingdata. The source of error include the lighting noise, hot electronicnoise, CCD function, lens distortion, quantification error, thedifferent between saving frame and CCD, temperature, vibration, andso on.