¿µ¹®³»¿ë (English Abstract) |
This paper presents a digit recognition method of analog meters that measure water, gas and electricity consumption. Digit appearance in images can be significantly affected by factors such as camera parameters and illumination changes, which make it difficult to recognize digits. The proposed method performs the following procedure; digit region extraction using histogram distribution, gradation correction of image, binarization with adaptive thresholding, non-digit region rejection, and optical character recognition. Experimental results show that the proposed method yields a low 3.5% recognition error rate.
|