Study On The Method Of Automatic Reading Of Water Meter

China is a country with a shortage of water resources,China's total freshwater resources rank fourth in the world.However,China's per capita water resources are only one-fourth of the world's average.Therefore,the rational use and management of water resources is a major issue related to the national economy and the people's livelihood.Most of the various automatic meter reading technologies that came into being there are still many deficiencies,and there is a large digital error rate that cannot be truly automated.For this reason,this paper proposes an improved water meter number recognition method.In this paper,after deeply researching the current research status and development direction of automated meter reading system,combined with the actual engineering project and the current image recognition technology,a digital display number recognition method based on SIFT algorithm is proposed and proved to meet the design requirements of the system by experiments.The main work of this article includes:(1)Starting from the preprocessing of images,the problem of color loss in the process of overexposure of photographed pictures and conversion of color pictures into grayscale images are firstly analyzed,for the pixels in the process of converting a color image into a grayscale image problem,a SIFT image matching algorithm based on image graying was proposed,which converts from RGB color space to LAB color space,and uses the color offset to increase the contrast of the grayscale image;for the overexposure problem,firstly,add the exposure offset to grayscale image and normalize the grayscale pixels.In the last,the improved feature matching algorithm based on SIFT is proposed.Experiments prove the effectiveness of this algorithm.(2)In the position and extraction of the water meter numbers,SIFT algorithm is used to match feature points and find "m3" signs.realize automatic positioning water meter indicator area combined with subsequent operation.the Canny operator is used to detect the edge of the water meter,and the number area to be identified by the projection method.The image is rotated and scale transformed by the affine transformation method for the image with different tilt and focal length.The characters are segmented by horizontal and vertical projections,and the segmented numbers are normalized.Experiments show that the method has good noise immunity and high recognition accuracy.(3)In character recognition,different recognition methods are proposed for semi character and whole character.The word character is identified by the standard template matching method;for the semi character,scanning the continuous upper and lower semi character,the value of the semi character is determined according to the identified upper and lower semi character and the matching coefficient of each template.The experiment shows that the algorithm has good recognition rate.