Research on Reading Methods for Pressure Gauges in Oil and Gas Fields Based on Machine Vision

Pressure gauges are widely used measuring instruments in the industrial field for measuring the pressure of liquids or gases. Traditional pressure gauges typically display pressure values using a pointer. However, in certain special situations such as high temperatures, high pressures, or harsh environmental conditions, manually reading the pointer's position can be difficult or risky. Therefore, in recent years, researchers have started exploring the use of computer vision technology to achieve automated pressure gauge pointer recognition. This paper employs the Otsu threshold segmentation preprocessing method to process acquired images and proposes an optimized recognition method that combines the Hough transform with convolutional neural networks for pressure gauge pointer recognition. The research results indicate that the recognition method, which combines the Hough transform with convolutional neural networks, can quickly determine the pointer's position, and the instrument readings identified by the algorithm are relatively close to the actual readings. Therefore, the proposed pressure gauge recognition method in this paper is effective.