Measurement Of Key Dimensions Of Large-size And Thin-wall Pieces Based On Vision Inspection Technology

Vision inspection technology has been developed quickly with the development of computer techniques. Compared with the contact-type measurement, vision measurement can reach the goal of measuring on-line, non-contact and quickly, which indicates a better future of its widespread applications in the industry.The purpose of this thesis is to develop a method to obtain the exact key dimensions in large-size and thin-walled pieces with vision measuring technology, especially for locating the border of two cylinders or cones as well as the position of the thread lines in workpieces. The main research work in this thesis includes the follows.1) The present vision measuring techniques are studied in details and their advantages and disadvantages are analyzed.2) The common structures of a vision measuring system are analyzed. The vision measuring system used in this project, including the experimental platform and transmission mechanisms are built. The prototype of the measuring device has been tested.3) Vision measuring techniques and image processing techniques are studied in detail. All kinds of environmental factors which have impact on the accuracy are discussed. Several image processing methods including median filtering and dual threshold filtering, and the center extraction methods of the linear structure light are investigated and compared.4) B-spline fitting and edge-likelihood estimation methods are firstly proposed to locate the shoulder height and thread position of workpieces. These algorithms are used to detect the features with high accuracy. Experiments have been carried out to prove the feasibility of the algorithms.5) Based on Tsai's calibration theory, the calibration software based on Visual C++ is accomplished. The results are obtained through the calibration experiments.6) The algorithms to deal with target image are proposed. The errors of object coordinates and image coordinates are obtained. A measuring accuracy estimation experiment using B-spline fitting and edge-likelihood estimation methods is carried out by measuring the gauge block surfaces. The possible error sources are analyzed as well.