Uncertainty of vision-based measurements in dynamic applications and the use of Cepstrum to improve the results

Nowadays, digital image correlation (DIC) is widely used in optical full field measurement. When dealing with dynamic conditions, measurement uncertainty becomes critical due to the motion blur caused by the motion. Therefore the present study aims to estimate motion blur and then remove motion blur to improve DIC uncertainty in dynamic conditions in the case of translating target. The study focuses on 2D DIC. In the case of 3D DIC similar problems will arise and therefore, a complete understanding of two-dimensional conditions will be of great help to further studies dealing with 3D conditions. The whole work can be divided in to three parts. In the first part, two methods using cepstrum analysis are proposed to estimate motion blur. The two methods are both applied to compare estimated value and to perform the error analysis respectively. The second part of the study aims to remove motion blur. A new motion blur removal technique is proposed. The basic idea is to remove the blur information in cepstrum domain, which has a de-blurring effect in space domain. In the third part, since we can estimate motion blur and remove motion blur, we can test in real DIC analysis to check the effectiveness of the motion blur removal technique proposed in previous two parts. Therefore, several reference image types are used in the test. The results show that uncertainty of DIC is reduced after motion blur removal by the technique we propose. So it proves that the introduced technique is an effective method for motion induced uncertainty reduction.