Data Acquisition Techniques In Color-Field Digital Photoelasticity

In automatic acquisition and procession of the photoelastic data, there are some troubles about how to acquire single isoclimcs and to improve the precision of evaluating isochromatic fringe orders in the thin slice of the three-dimension model when the actual problems are solved by photoelasticity. Because color isochromatic fringes appear in the stress model based on a plane polariscope in white light, we studied the problems systematically by a digital camera as image inputting device, digital image processing technique and BP artificial neural network.Based on the work that has been reported before, a technique named five-step color phase shifting technique (FCPST) to acquire single isoclinics is presented. Relevant theory is derived and explicit conditions for directly determining the isoclinic parameter in the range of [0- 3i /2] are given. A color camera records five isoclinic images coupled with isochromatics from a plane polariscope with five different settings in white light, respectively. Computer simulations and experiments have been carried out with a circular disk under diametral compression and complex two-dimension and three-dimension engineering application problems. The results prove that the FCPST is better than the other phase shifting techniques (PST).Three main error sources are the variation of intensities, the orientation errors associated with the configurations of the polariscope and the initial stress existing in the model. A quantitative method is presented to analyze and estimate the influence of different error sources on the final isoclinic results calculated by FCPST or the other PST. Some guidelines and measures to control and decrease the influence of the errors on the results are given.An automatic order determination of color isochromatic fringe based on BP artificial neural network is presented. The complex function between isochromatic fringe orders and RGB gray values of the image pixels is described by BP artificial neural network. The technique can improve the precision of experimental stress analysis in three-dimensional photoelastic thin slices.In order to reduce photoelastic data from the digitized fringe image and to avoid the repetitive work, a photoelastic digital imaging software independent device is presented. Thesoftware has realized the techniques presented before and has other functions, such as the simulation of a disk under diametral compression, various error analysis, phase unwrapping, real-time display of image information, and so on.