Identification Method Of The Digital Readout Meter Header

With the development and popularization of science and technology, the management means of every walk of life is changing to automation or semiautomation. Contraposing to the various defects of the traditional manual water meter reading, people bring forward an automatic water meter reading technology, and digit character recognition is a very important link, which is the content of this paper.The water meter image in this paper is shooting by digital cameras, recognition process including: image preprocessing, tilt correction, character segmentation and character recognition. Image preprocessing including image denoising and image binarization. Against unknown light conditions and classic Bemsen algorithm exists fake shadow, using LEVBB algorithm binarization, and getting a good result. Image tilt correction uses Hough transform to extract water meter frame line, and the results are statistical average, and then calculate the tilt angle. Finally, the image can be corrected by affine transform. Character segmentation first use of a priori knowledge thick segmentalizes the image to be five characters in the general location. Then by scanning to wipe off black-frame, Opening operation denoising, using connected component delete large area of smudge further process the image. Finally using of projection segmentation, can find the exact location of the character. Character recognition including entire character recognition and half character recognition two parts. The entire character use template matching to identify which is according to Hamming distance. The half character recognition use template matching based on feature extraction to identify and both of them obtain a high rate of correct identification.