Font Size: a A A

Sub-pixel Displacement Algorithm Based On Image Correlation Method And Experiment Study

Posted on:2010-04-02Degree:MasterType:Thesis
Country:ChinaCandidate:W P WuFull Text:PDF
GTID:2178330338485112Subject:Solid mechanics
Abstract/Summary:PDF Full Text Request
Digital image correlation method (DICM) has played a decisive role in the field of mechanical testing, as a kind of precise optical measurement method by computer-aided testing. The key to DICM is the calculation precision and stability of the algorithm of sub-pixel location.Firstly, in this paper, the domestic and overseas research progress and application are introduced; after the introduction of the principle of DICM, the three steps search (TSS) method based on motion estimation of video image is introduced to correlation search of integer pixel in DICM, and the paper presents the algorithm of generation of computer simulative speckle images.Secondly, the paper summarizes the algorithms of image sub-pixel displacement location, with emphasis on image interpolation algorithm,correlation coefficient interpolation or fitting algorithm, gradient-based algorithm(based on image gray or correlation functions), these three kinds of algorithms have been used most recently, the excellence of the algorithms are compared by using simulative speckle images.In succession, on the basis of the summary of the algorithms of image sub-pixel location, a novel algorithm based on moving least squares curved surface fitting of correlation coefficients is proposed, which combines the moving least squares principle with curved surface fitting, and then the paper optimizes the new algorithm. The paper also puts forward the optimal subset size of correlation calculation, the influence radius of weight function and the orders of the fitted curved surface. The novel algorithm is compared with the conventional algorithms of curved surface fitting of correlation coefficients and gradient-based algorithm, especially in the aspect of noise-resistance characters.Finally, the paper applies the novel algorithm to measure the deflections of the equal strength beam. The paper describes the way of characterization operation of surfaces of the experimental specimen. In the aspect of calibration of the experiment, two methods are presented, one is straight line fitting based on least-square method; the other is circle fitting, which is also based on least-square method. Based on the software platform of Matlab, two kinds of software have been developed, one based on algorithm of gradient, the other based on algorithm of moving least squares curved surface fitting. The two software is applied to the measurement of the deflections of the equal strength beam. Comparing the calculated values with the theoretical values, the results are satisfying.
Keywords/Search Tags:Digital image correlation method(DICM), Sub-pixel location, Curved surface fitting, Moving least squares(MLS), Influence radius, Experimental calibration
PDF Full Text Request
Related items