Research And Development Of Video Extensometer Based On Machine Vision

With the rapid development of China's industry, the types of new materials are emerging. The merits of the material properties determine the quality of the product, and how to detect the high accuracy of the material properties is an urgent problem to be solved. The extensometer is one of the basic devices to measure properties of materials in present stage, and to improve the detection accuracy and stability of extensometer is getting more and more attention.In view of the domestic and foreign video extensometer of the status quo, this paper designed a set of non-contact video extensometer. Its feature was the design of the three types of identification signs and related identification algorithm, and it can be suitable for all kinds of test pieces of the tensile test, meanwhile,it have been greatly improved in recognition speed and accuracy. The important practical significance was that: Firstly, it can replace imported equipment in performance. Secondly, it provided a reliable displacement measuring system for the tensile testing machine, and laid the foundation for the automatic tensile testing machine. Thirdly, the range of its adaptability was strong, and it can replace the electronic extensometer to save the production cost. In terms of hardware, according to the practical application and the effect of light on the image acquisition, the LED lighting system was designed. Considering a lot of comprehensive factors, the camera was selected and for the different range of test pieces, ordinary lens, lens and telecentric lens were configured. In terms of software, the software function was divided into image acquisition, image processing and data storage module, and it reported output based on Excel.Because the video extensometer obtained the deformation of the specimen by marking on the specimen and measured the change of position after the marking, the used mark method, extraction of the target image determined the video extensometer measurement precision. In order to avoid the fixture interfere to sign recognition, this paper used the bar code to extract marker region. After filtering the extracted image, the sub pixel interpolation subdivision was carried out, which made the detection accuracy in the submicron level.In the way of marking, it designed three kinds of different marks, circular, linear, and two straight line segments. Edge detection was used to identify the edge of the circle for circular marks, and for the discontinuous points, it used Hough transform circle detection method to convert it to the parameter space of the accumulator unit vote. The parameters corresponding to the maximum voting value was the center coordinates.For the sign of the straight line type, the detection method was to transform the image space and the parameter space, to vote for the accumulator unit for every point, to vote the statistical distribution for the window reached threshold. The curve interpolation was used to calculate the angle and the pole diameter, and the straight line parameters were determined. For the type of two straight line, the line intersection point P was detected as the point of displacement.The research results of this paper have been verified in the experiment site. The image was collected by the camera, and after processing the image, the deformation and the stress and strain curves were drawn. Through the verification of the test results, it was concluded that the system can meet the accuracy requirements, and the operation was stable and reliable.