Strain Measurement Of Sheet Metal Parts Based On Multiple Grid Images Taken By One Single Hand-held Digital Camera

With high speed development in manufacturing, the traditional manual strain measurement methods of using mylar tape and tool microscope have gradually exposed some disadvantages such as low efficiency and poor accuracy in most sheet metal stamping situations. In order to overcome various limitations of the manual strain measurement methods and meet the requirements of strain measurement in the filed of sheet metal stamping, a non-contact strain measurement method, which is based on multiple images taken by one single hand-held digital camera, is realized in this theses. More specifically, a square grid pattern is printed on the surface of the sheet metal in advance, and the 3D deformation of the grid after the stamping process is calculated by analyzing the multiple images. The strain distribution of the sheet metal product is further calculated according to the 3D grid deformation.The key techniques of the method have been deeply studied in this thesis, including grid nodes extraction, camera position and orientation calibration, matching and reconstruction of grid nodes, grid strain calculation, etc. The main research contents and innovative points are summarized as follow:①First of all, an overall scheme for grid nodes extraction is designed. And then, some problems are figured out in preprocessing stages such as image binarization and thinning, and corresponding solutions are carried out; Finally, a coarse-to-fine algorithm based on the thinning image of grid lines is proposed, so that the sub-pixel image coordinates of grid nodes are precisely extracted and the topological relationship of each node is established.②On the basis of deep research on the theory of multiple view geometry in computer vision, a camera position and orientation calibration approach is taken into practice.③The limitations are analyzed when using the classical region-based and feature-based matching methods into the grid image matching, and a matching algorithm based on the grid topological relationship is proposed. Regarding to the reconstruction problem of more than two corresponding image points, a specific solution is proposed through the theoretical derivation.④A grid strain calculating method is determined in terms of grid strain analysis theory and specific application.A prototype strain measurement system, namely MOSAS(Monocular Strain Analysis System) is independently designed and realized based on the above key algorithms and techniques. The performance of the system is strictly verified. Experimental results show that, MOSAS system has high automation, good precision and stability, as well as ability to meet the requirements for measuring the strains of complex and large sheet metal parts.The research achievements in this thesis have successfully solved the problems which exist in the contact strain measurement methods, and effectively meet the high efficiency and good precision requirements of strain measurement in the field of sheet metal stamping. Based on this research, reliable strain data can be provided for subsequent sheet metal formability analysis.