• Home
• Preliminary Program
• Authors center
• Important dates
• Keynote speakers
• Sponsors and exhibitors
• Scientific Committees
• Conference venue
• Travel information
• Registration
• Contact

Tweets by @PhotoMech2015

11:00   Digital Image Correlation Algorithms Chair: Bertrand Wattrisse 11:00 20 mins A NEW DIGITAL IMAGES CORRELATION ALGORITHM TO GET RID OF LIGHT REFLECTION PROBLEMS Martin Poncelet, Hugo Leclerc Abstract: Light reflexion frequently occurs on specimens experiments instrumented with cameras as soon as their shape is not flat. In this case, kinematic full-field measurements performed with the acquired images are usually spoiled. To prevent this, a new experimental setup is proposed, based on the idea of using several alternated lighting sources. Depending on where the reflexions are on each image, a Digital Images Correlation algorithm weights the images regions to enhance results quality. 11:20 20 mins RANDOM DISPLACEMENT ERROR ASSESSMENT OF SUBSET-BASED 2D-DIC Yueqi Wang, Pascal Lava, Dimitri Debruyne Abstract: This article concerns the theoretical assessment of the random displacement error measured by subset-based 2D-DIC. According to the algorithm of DIC, derivations are carried out to provide analytical solutions to estimate the random measurement error. This work excludes simplifications in the present solutions from literature, so that a more general error estimation is obtained. 11:40 20 mins AUTOMATIC MESH REFINEMENT IN GLOBAL DIGITAL IMAGE CORRELATION Lukas Wittevrongel, Dimitri Debruyne, Pascal Lava, Stepan Lomov Abstract: A full automatic refined global DIC algorithm is presented. It is based on the combination of a novel adaptive global DIC algorithm containing higher order p-elements and convergence of element energy (equivalent L2 norm). Both principles are combined in the correlation platform ‘AdaptID’, housing the automatically refined global DIC approach. Using the global approach insures a continuous displacement field while using higher order elements result in large elements with a good spatial resolution and noise robustness. The implementation of hierarchical p-elements leads to the possibility of adapting element during correlation. Automatic refinement is obtained by using convergence in element strain/displacement norms. In the proposed method, an element keeps being refined until its L2-norm (strain or displacement) converges. For the mesh, refinement continues until all elements are converged. In this way each element has its own most optimal polynomial order. Validation of the proposed algorithm showed great performance in terms of resolutions and robustness to experimental influences. An extra major benefit is the automatic mesh refinement which rules out most user dependency. 12:00 20 mins AN EXPERIMENTAL STUDY OF CAMERA SENSOR NOISE PROPAGATION TO DISPLACEMENT MAPS OBTAINED BY DIC Benoît Blaysat, Michel Grédiac, Frédéric Sur Abstract: The predictive formula of resolution displacement maps obtained using Digital Image Correlation tools are revisited and validated in this paper. Indeed, the usual formula available in the literature does not take into account the DIC required sub-pixel interpolation. Moreover, a noiseless DIC tangent operator is defined to also minimise noise propagation from images to displacement maps. These points are now taken into account in the proposed update of the predictive formula. The experimental validation is then done, it is shown that specific image pre-processing tools are required. Usual raw images are indeed corrupted by noise and consequently do not satisfy assumptions needed for predictive formula elaboration. Thanks to the pre-process presented here, image noise properties are corrected, thus assumptions of the predictive formula are satisfied and the prediction is validated. 12:20 20 mins COMPARISON OF FINITE ELEMENT MODEL UPDATING AND INTEGRATED DIGITAL IMAGE CORRELATION FOR PARAMETER IDENTIFICATION Andre Ruybalid, Johan Hoefnagels, Olaf van der Sluis, Marc Geers Abstract: To improve the reliability of multilayered microelectronic systems, predictive models are needed, which require material and interface parameters. In order to adequately identify model parameters, full-field identification methods are suggested. At the heart of such a method lies Digital Image Correlation, which captures full-field kinematics, exhibiting extensive information about material response. The recently developed, one-step method of Integrated Digital Image Correlation (IDIC) was compared to the conventional, two-step method of Finite Element Model Updating (FEMU). Despite the mathematical resemblance of both methods, it is concluded that IDIC is less laborious than FEMU, because of its one-step nature, and more importantly, produces more reliable identification results, because it was found to be less sensitive to image-noise, contains less systematic error sources, and is not subjected to critical choices that affect the accuracy of the identification results. 12:40 20 mins A PARALLEL APPROACH FOR FAST AND DENSE STEREO-CORRELATION Guy Le Besnerais, Yves Le Sant, David Lévêque Abstract: We present a new approach for fast estimation of dense 3D displacement fields from stereo images. It is based on a previously published correlation algorithm which is ideally suited for parallel processors such as GPU. We describe the algorithm and its application to stereo-correlation and present results on real experiment. The uncertainty appears comparable with previously proposed softwares, with a much better computational performance.