Image Motion Blur Mechanism-Based Measurement Method for Low-Frequency Vibration Amplitude and Direction

low-frequency vibration measurement is crucial to the mechanical system dynamic performance testing, structural health monitoring, and auxiliary medical diagnostics. Machine vision (MV) has gradually been applied in low-frequency vibration measurement because of its portability, flexibility, and noncontact nature. However, the relative motion between the camera and the measured vibration object inevitably causes motion blur in the image, which decreases the measurement accuracy. To eliminate the influence of motion blur on vibration measurement performance, this study proposes a new motion blur mechanism-based low-frequency vibration measurement method. This method models the mathematical relation between stripe features and vibration parameters in the fast Fourier transform (FFT) spectrum so that the vibration amplitude and direction can be solved from the stripe width and direction. Meanwhile, median filtering and Kaiser window are utilized to preprocess the vibration-blurred images to reduce environmental noise interference and spectrum leakage and to improve the contrast of the stripes. Additionally, the Canny edge detection method is added to Radon transform to overcome its limitation in direction detection, so as to improve the accuracy and reliability. To avoid the problem of excessive edge smoothing in stripe edge extraction, the method improves Canny edge detection to achieve accurate localization of spectrum stripe edges. The simulation results show that the maximum relative error in amplitude was 1.7% and the maximum difference in direction was 0.19°, respectively. Experimental results from the 1–7 Hz vibration measurements indicate that the proposed method achieves a maximum relative deviation of 1.8% in amplitude and a maximum difference of 0.04° in direction.