ROTATIONAL SPEED MEASUREMENT THROUGH LOW-COST IMAGING DEVICE AND DIGITAL IMAGE PROCESSING

Abstract

Rotation speed detection is a fundamental parameter for system monitoring, process control, and fault diagnosis of rotating machinery. The objective of this work is to develop a system for estimating rotation speed through low-cost optical sensors, such as cameras or webcams, and digital image processing algorithms. The system presented is based on the acquisition of a video taken from the rotation of the rotor through a camera and its processing on a computer. Two types of videos were used: one showing a simple mark made to the rotor cross-section and the other the inhomogeneous side surface of the rotor. In the video taken, a pre-processing stage was used to improve the qualities of the images, a stage to calculate the level of similarity between the first image and the following, and the spectral analysis to determine the rotation frequency from Fourier’s Transform. The performance and effectiveness of the system are verified through experiments using videos made in virtual and real systems where the frequencies used were correctly obtained. It was found that in the simulation work, the robustness of the system to different types of noise depended on the pattern presented by the rotor. While for real videos, it showed a correct and precise performance for high rotation speeds, but for low ones, high relative errors were obtained by coinciding the real value with the order of resolution of the tachometer.