Research on the Methods to Reading Pointer Instrument Based on Machine Vision

This paper puts forward to a novel intelligent methods to reading point instrument to replace human reading. The images of pointer instrument are captured. Firstly a segmentation algorithm based on unsupervised learning is presented to segment the image for locating the scales and pointer. This algorithm combines the fuzzy C-means method with fuzzy support vector machines, and can subdivide the image into the objects and background efficiently. Secondly the according to the characteristic of pointer instrument, the central of instrument dial is confirmed by Hough transform. And then the central angles of the scale regions and pointer region are calculated. At last the reading value of pointer instrument is determined. The Experiments prove the method presented by this paper is effective.