Research on Image Recognition Algorithm for a New Type of Intelligent Ammeter System

Abstract

In order to solving the problem of the image skew, partial character and decimal fraction calculation that effect the computation accuracy and speed. This article put forward a new algorithm including meter digital image preprocessing and digital identification. Preprocessing algorithm is based on texture feature to position roughly at first, then using the morphological method to position it accurately, in the end, using Hough transform and Wavelet analysis to correct image and remove the interference and separate the single digital from the complex background. In the Digital identification, it uses the different algorithms to the whole and the partial digital image. The whole digital image uses the Wavelet analysis and another uses the digital recognition algorithm which is based on pixel and scale. To verify the algorithm by the 800 meter image, the veracious rate is 99.6% for single digit recognition, the veracious rate is 98.1% for whole meter image recognition, and the speed is double faster more than using traditional algorithm. The results show that the algorithm has great practical value.