Study On The Displacement Field Measurement Algorithm Based On Digital Image Correlation

With the development of experimental mechanics,instrument science and other disciplines,the full field displacement measurement method based on digital image correlation is developing rapidly,which has been widely concerned for its advantages of simple optical path,good environmental adaptability,wide measurement range and high level of automation.Although the accuracy of the current digital image correlation method for displacement field measurement has reached the standard of engineering measurement,but the efficiency and success rate of the integer-pixel displacement search are still insufficient,especially in the higher precision displacement measurement,which involves sub-pixel displacement calculation,and the gray value at the sub-pixel position needs to be interpolated and reconstructed,so the calculation efficiency has become the bottleneck of the application of digital image correlation method.In this paper,improved integer-pixel and sub-pixel displacement calculation algorithms are proposed,and an efficient displacement field measurement algorithm without interpolation is formed by combining the two methods.The main work is as follows:(1)An integer-pixel displacement search algorithm based on improved particle swarm optimization is proposed.For the case of small displacement,a set of predetermined particles near the center of the search element is introduced,which reduces the number of iterations of the standard particle swarm optimization algorithm and the search time.After the iteration,the three-step method is used for fine search,which solves the problem that the particle swarm optimization algorithm is easy to premature,and ensures that the searched points are the peak points of correlation coefficient,and improves the success rate of the integer-pixel displacement search.Compared with the unimproved algorithm,the search time is reduced by 12%,and the success rate is increased to nearly 100%.(2)A gradient method based on neural network error compensation is proposed,which can get sub-pixel displacement without interpolation.The nonlinear inherent error of gradient method is analyzed.The three-layer structure BP neural network is trained for the measurement error of gradient method,so as to realize the prediction and compensation of calculation error.After compensation,the mean value of error is reduced by about 50%,and the standard deviation is reduced by about 30%.Comparing the improved gradient method with the classical Newton Raphson iterative algorithm(N-R algorithm for short),the results show that the accuracy of the two methods is equal,while the former is three times faster than the latter.(3)The simulated speckle is adopted to further study and verify the displacement field measurement algorithm proposed in this paper.When the size of subset is 21 × 21,the SNR is 25 d B,and the number of speckles is above 1000,the calculation accuracy and efficiency are the highest.During the measurement of the simulated speckle displacement field,it is found that when the integer-pixel and sub-pixel algorithms are combined although the accuracy of the calculation speed decreases under the influence of noise,it is still equivalent to the displacement field measurement method based on the NR algorithm.And its computation speed is still faster than the latter.(4)The human-computer interface based on Matlab GUI is designed,and the algorithm proposed in this paper is verified by rigid body translation experiment and sheet metal uniaxial tensile experiment.The experimental results show that the difference between the proposed algorithm and the N-R algorithm is not significant,and the measurement errors are within the acceptable range,which verifies the reliability and practicability of the proposed algorithm.