Design and Realization of Reading Recognition System for Sound Level Meter

This paper presents the necessity of reading recognition in calibration process of sound level meter, and analysis the problems of existing display reading algorithm, then presents a novel reading recognition system for sound level meter. The system find the position of number string by image difference and connective region analysis, and pre-process the image by local thresholding based on neighborhood averaging method, then recognizes the number by pattern grid and template matching. The system has the characteristics of rapid and accurate. The experiment shows the system is well competent for calibration work of sound level meter.