Research On Improved Sub-pixel Extraction Algorithm Based On Online Detection

As a measurement method combining traditional optical measurement technology with modern digital image processing technology,digital image correlation method is often used in displacement or strain measurement.At present,the main research trend of sub-pixel digital image correlation methods focuses on the calculation efficiency and measurement accuracy.It is particularly important to improve the calculation efficiency on the basis of ensuring the accuracy of displacement/strain measurement.Therefore,this paper mainly studies how to further improve the search speed of sub-pixel,and discusses the advantages and disadvantages of common displacement detection algorithms,so as to better meet the needs of actual engineering measurement.The topic selection is based on the cooperation project of Anhui Institute of Metrology Research and the campus-industry cooperation industry innovation guidance fund project.this paper studies a subpixel extraction algorithm of geometric center of gravity method,and improves the calculation time and anti-interference ability of the commonly used subpixel algorithm,aiming to explore an online measurement technique with simple algorithm content and high calculation accuracy.The main research contents of this paper are as follows:First,the basic theory of digital image related methods is described,the limitations of existing engineering measurement methods are summarized,the convenience and research status of current digital image related methods are discussed,and displacements are proposed based on several commonly used subpixel displacement algorithm theories and detection methods / Strain measurement problems.Compare commonly used sub-pixel measurement algorithms such as surface fitting algorithm and gradient algorithm.Introduce the existing binary primary surface fitting,binary quadric surface fitting and Gaussian surface fitting formulas,and gradient operators commonly used in the gradient method.On this basis,a new sub-pixel search algorithm geometric centroid method is proposed,and the basic principle and program steps of the method are described in detail.The existing sub-pixel correlation coefficient fitting formula is carried out through a simple addition and subtraction algorithm.To simplify,and add the index value suitable for the measurement of different speckle surfaces in the program,at the same time,the method of reducing the search matrix is used to reduce the displacement measurement error and improve the calculation efficiency.Then,the grating reference measurement system is introduced,the construction of experimental devices such as micro-displacement calculation and speckle image capture,and the system calibration of the calculation results are introduced.Discuss the measurement results of the above algorithms in different background lights,different index values and different materials,and verify the calculation accuracy of the geometric center of gravity method.At the same time,through the calculation time of the algorithm,the geometric centroid sub-pixel extraction algorithm is compared with the commonly used digital image sub-pixel displacement algorithm,and the type of error that may exist in the displacement experiment and the way to reduce the error are introduced.At the same time,sub-pixel calibration experiments are carried out to verify the measurement accuracy of the geometric centroid method in the standard image displacement.Finally,comprehensive experiments verify the advantages of the geometric centroid method in terms of calculation accuracy,calculation efficiency,and anti-interference.The algorithm can meet the online measurement of micro-displacement with a certain accuracy,and has high practicality in practical engineering detection.