An Automatic Meter Reading Method Based on One-dimensional Measuring Curve Mapping

Abstract

Take advantage of the good matching performance of scale-invariant feature transform (SIFT) features between multiple images about a textured object, our automatic meter reading method calculates a homography matrix H between the template image and the test image, uses random sample consensus (RANSAC) to filter out false matches, ensuring an accurate H, and finally maps the measuring curve from the template image to the test image. On the path of this measuring curve, the relationship between the pointer and the scale is calculated based on the gray scale change in the image. The pointer is detected and the reading can then be obtained using a particular formula. Our method provides rotation invariance, scale invariance and higher reading accuracy. Experimental results show the effectiveness of our method in the substation environment.