The Methods And Techniques Of Deformation Measurement Based On Machine Vision

With the development of domestic industry, the requirements to the new materials performance are also increasing. Among the requirements, the stress and strain performance test are indispensable. In the strain measurement, the traditional contact measurement contain many shortcomings: such as complex operation, unable to adapt to the intelligent requirements, high environmental requirement which is not suitable for high temperature and high pressure measurements. With the advantages of wide applicability and intelligence, the non-contact measurement based on machine vision is becoming a hot research spot in the area of deformation measurement. In this paper, starting with the aspects of software algorithm,the improved image processing algorithm is used to improve the accuracy of deformation measurement.Firstly, the improved wavelet algorithm is introduced in the application of image filtering and image edge detection. The process of image acquisition will be affected by the internal and external factors such as unstable lighting, the change of the temperature sensor. These factors are likely lead the image to obtain noises. So the image filter algorithm is used to eliminate the noises. In this paper, an improved wavelet threshold algorithm is adopted to image filtering. By taking advantges of wavelet multi-scale feature, and the image is decomposed to find and remove the wavelet coefficients of noise, then reconstruct the image to complete the filter. Next the fixed threshold method is used to binarize the grey value of image. The image after binarization only exists two kinds of grey value, thereby it can reduce the calculation of edge detection. On this basis, the wavelet algorithm is used for image edge detection. Through comprehensive comparison of wavelet algorithm with the traditional algorithm, the conclusion is that the wavelet algorithm is better in the area of edge recognition accuracy and computing speed, so as to get the optimum algorithm of edge detection. At this time, I complete the positioning of image edge in the pixel level.Secondly, the subpixel edge location is introduced in the application of the deformation measurement based on machine vision. In view of the subpixel edge detection, the gray moment algorithm and least squares algorithm are used for research. By linear fitting the subpixel location line to get two tag lines. Then By measuring the image pixel distancebetween the two tag line, the specimen deformation is calculated according to the camera's calibration value. Studies show that the effect of subpixel location using least square method and linear fitting method to improve the detection accuracy is effective.The deformation measurement system runs under the Windows XP system, by using C++language combined with OpenCV open-source database to complete the interface design and software programming. The system realizes the target of specimen deformation measurement and control as one with faster speed and higher accuracy. In addition, this system program adopts the modular design, strong portability, which is easy to combine with the control program and apply to the practical industrial application. In the experiment, the testing machine is used to tensile specimen, choose industrial digital camera to capture the images.Through related data obtained by image processing, the test method data is used for calibration, so as to complete the precision measurement of specimen deformation.