Research On Sub-Pixel Accuracy Algorithm Of Digital Image Frequency Domain Method

Modern optical measurement technology has been widely used in many areas because of the advantages of no damage,non-contact,full-field measurement,no additional quality and high resolution,etc.The Digital Image Frequency Domain(DIFD)method is especially appropriate for the practical engineering measurement due to its simple facility,high resolution and low requirement to the environment.However,the existing DIFD method still has ample room for improvement in the resolution of sub-pixel and computing time.In this paper,some key problems are deeply researched and new methods are proposed to improve the sub-pixel precision and the computing efficiency.Based on the DIFD method,several methods are proposed to calculate the sub-pixel precision displacement in this paper.According to the up-sampling technology,the surface fitting iteration method and Newton iteration method are analyzed to get the displacement generated by simulated digital speckle images,and result shows that the two methods have the same accuracy while surface fitting iteration method is more efficient.The accuracy of the result and the calculation time are affected by various factors while using the DIFD method to calculate the sub-pixel displacement.The influence of the size and the number of the speckle particles is discussed in order to get the best quality of the simulated digital speckle images.The noise robust is also verified by adding Gaussian noise to the images.The computing time and sub-pixel accuracy are mainly determined by the calculating size.By trying different calculating size,the best one is selected to suit both the efficiency and the accuracy.The images with higher quantization level are proved to be able to get more accurate displacement and less error.The DIFD method and the traditional Digital Image Spatial Domain(DISD)method are compared in terms of the accuracy and the efficiency.Result shows that the DIFD method owns more advantages than the DISD method.The rigid body translation experiment verifies the feasibility and accuracy of the DIFD method by calculating the real speckle images.Other two experiments are also designed to prove that the DIFD method is fit for engineering measurement.