Abstract: In order to improve the accuracy of automatic reading of tilted meter images，a novel algorithm

based on ORB and improved region growing method was presented. Firstly，the ORB algorithm was applied to detect the feature points of the target image and the template image，and the transformation matrix used to achieve tilt correction was calculated according to the matched feature points，the perspective transform was used to realize tilt correction. Then，the least square method was applied to detect the center of the template image，and the center after the inverse transform was used as the seed point. The gray scale adaptive threshold based on the pointer area was applied to detect the pointer area as the growth criterion. Finally，the projection method and Hough transform were adopted to extract the pointer rotation angle，and the meter reading was calculated according to the angle method.The experimental results suggest that the maximum relative error and the average consumed time of the proposed method are 2.96 % and 400 ms，respectively. Reading accuracy and calculation efficiency of the proposed approach can fully satisfy actual requirements for automatic reading of the analog meter on the spot. Therefore，the proposed method has a positive effect on boosting the construction of energy internet of things and digital power grids.

      Key words: image processing；tilted meter；automatic reading；pointer extraction；region growing algorithm