Camera-Based Vibration Sensor Using Time-Directional Division-Based Motion Estimation Without Unwrapping

Video cameras can gather spatial data of high density, given that each pixel functions as a sensor. Visual vibration sensor using video stands out in the engineering measurement field with the rapid development of digital sensors. However, computationally intensive phase unwrapping or high-speed acquisition has to be employed to remove the intrinsic phase. As a result, efficient manner and high fidelity are still challenging for camera sensors with large amount of data and complex signal components. Here, a novel camera-sensor-based motion estimation method using time-directional division is proposed to solve the aforementioned issues. After obtaining the frequency features of each frame of the motion video using the 2-D Gabor transform, division operations are directly applied to the frequency features of the images from one to the subsequent. In this case, phase variations are therefore obtained from the corresponding analytical signal without performing the phase unwrapping and/or high-speed acquisition, as well as removing the intrinsic phase. A series of simulation and experimental tests are conducted to prove the feasibility and effectiveness of the proposed method, and the results indicate that the proposed method can demonstrate a significant improvement in terms of accuracy and time efficiency when compared to the state-of-the-art phase-based motion estimation (PME) methods with and without phase unwrapping.