A Realization Of The Real-time Measurement For The Distortion Based On Machine Vision

With the rapid development of material technology, a variety of new materials appeared,and has been widely used in related research areas. The performance of the material is themain factor affecting the quality of products, therefore, an accurate measurement of thematerial performance becomes particularly important. In the measurement for the deformationof the material, the strain measurement is an important part in the tensile test of the material.Compared with the traditional contact strain measurement, strain measurement based onmachine vision is of high precision, non-contact and so on. In this paper, making the tensiledeformation of metallic materials as the background, and analyzing the requirements of thetensile properties for the metal material, a method of strain measurement for the distortionbased on machine vision is presented and further studied. Meanwhile a set of real-timedistortion measurement system of the material based on machine vision is developed.This article firstly introduces the research status at home and abroad about the real-timemeasurement of deformation based on machine vision, and determines the overall researchplan of deformation measurement system, including hardware design and software design ofthe measurement system. We get the image data steadily and fast by the Basler’s CCD camera.Then this article gets the suitable algorithm of image for this system by improving the basicmethods of digital image processing, thus it makes the strain of specimen is measuredaccurately.In this paper, the image processing methods includes image filtering, binary image, edgedetection, subpixel edge detection, line fitting of edge and so on, then edge detection and thesubpixel edge detection algorithms are important parts of this paper. Since the images havenoise, so in order to eliminate the noise in the image data, we should smooth the image thatwe obtained. Then make binarization processing to the image after smoothing, so we can getthe binary image. On this basis, the line profile of the specimens is isolated from the figure bythe edge detection; this paper chooses the improved Canny operator as the edge detectionmethod of the pixel level and makes image edge extraction in the pixel level. At this point, thecoarse location of image edge in pixel level has been realized. Then, this paper uses thesubpixel edge detection technique based on interpolation method to make subpixel accurate positioning of the line profile figure obtained which improves the accuracy of measurementeffectively without increasing the cost of hardware. After the image edge subpixel subdivision,we fit the discrete points of the edge by the linear least squares linear regression method, andwe can get a linear equation, thus we can get the pixel distance between the two marks.Finally, we can use obtained the pixel number multiply the calibration coefficients to get theactual size of specimen, and the article makes the error analysis for the experiment results.Windows XP operating system is the application platform for the system of deformationmeasurement based on machine vision. This paper uses Visual Studio2010as developmenttools. The interface of the system is developed and designed by configuring the Qt5.0. Thealgorithm of image processing is developed and designed by the Opencv open-source libraryof machine vision. And the program of system is developed and designed by the method ofobject-oriented. The software of measurement system achieves non-contact real-timemeasurement of the metal deformation in the tensile test basically, combining with the controlsoftware achieves the function, which realizing the system with integration of deformationmeasurement in real-time and control. and can be used in the real industrial measurement andcontrol system.