High Precision Videometrics Method Based On High Quality Image Acquisition

The measurement resolution of Videometrics is still difficult to meet the demand for small deformation measurement.An important reason is that the quality of the captured image from an experiment is not high enough,and the suppression of noise is not taken into account particularly in the image analysis algorithm.In response to this situation,this thesis proposes a new method,aiming at improving the image quality from both the image acquisition process and the algorithm process,thereby high-precision measurement can be achieved.In the image acquisition process,firstly,two important metrics,known as resolution and signal-to-noise ratio(SNR),are applied to assess the quality of image in Photogrammetry,and the related calibration method is also proposed.Accordingly,the influence on image quality caused by chromatic imaging and high-speed imaging modes and different imaging settings such as exposure time,gain,bit depth of A/D conversion of cameras,and aperture and focal length of the lens is investigated by experiments,and the related observations are summarized.Secondly,the effect of image quality on the measurement accuracy of feature-matching-based method and feature-registration-based method is simulated.The results show that,compared to the monochromatic imaging,the error of Corner Detection caused by the reduction of spatial resolution with chromatic imaging will increase by 2-3 orders of magnitude,and the error of displacement measurement of Digital Image Correlation(DIC)will increase by 3-5 times.In addition,it is confirmed that the influence of temporal noise and fixed pattern noise on measurement accuracy is different.Finally,the experimental results,which is consistent with the conculsions from numerical simultions,show that the imaging mode and imaging settings will affect the accuracy of Videometrics,and quantitative conclusions are achieved.It is found that the error is negatively correlated to image resolution and SNR.Specifically,the error of DIC displacement measurement caused by high-speed imaging will increase by one order of magnitude compared with that of standard imaging,and the amplitude of error will increase by different degrees if improper imaging settings are used.Therefore,the reference of a reasonable choice of imaging modes and imaging settings for the experimenter is provided.In the algorithm process,by extending and improving the Spatial-Temporal Subset based DIC,a low-SNR based image deformation analysis method is developed,and therefore higher measurement accuracy is achieved.Firstly,the influerence of the size of tempotal subset on the computational precision and computational cost is investigated,and the optimal tempotal subset is reccommanded.Secondly,the algorithm is extended to the image deformation analysis based on mark point recognition,which improves the measurement accuracy of displacement and strain effectively.What's more,the measurement error of the Spatial-Temporal Subset DIC based on the zero-order spatial shape function will further decrease to 50% compared with the first-order spatial shape function.Finally,two low SNR image analysis algorithms based on temporally nonuniform sampling and second-order temporally continuous deformation are developed respectively for the casewhen the sampling rate of the camera is not constant and the case when the sample rate does not match with the camera sampling rate.The results show that compared with the traditional Spatial-Temporal Subset based DIC,the developed methods can further improve the measurement accuracy of Videometrics.In addition,the error will further decrease if the size of the temporal subset increases.