Character Segmentation of Digital-Display Temperature and Humidity Instrument Based on Contour Features and Structural Rules

Abstract

In order to solve the problems of different types of digital-display temperature and humidity instrument segmentation such as imprecision, light influence and low generality, this paper proposes a method based on contour features and structural rules. Firstly, use the Canny edge detection method to extract the display screen, and utilize straight-line correction algorithm to do the screen correction. Then, characters are coarsely segmented based on the contour features and the different gray values between characters and background. Finally, accurate segmentation is done according to vertical projection and character structure rules. In this paper, the character segmentation is carried out with different types of digital-display temperature and humidity instruments. The experimental results show that the method can accurately segment the effective characters on different types of instruments.