点击次数：

论文类型： 期刊论文

发表时间： 2015-12-01

发表刊物： OPTIK

收录刊物： SCIE、EI

卷号： 126

期号： 24

页面范围： 5527-5533

ISSN号： 0030-4026

关键字： Machine vision; Dimension measurement; SNR; Large hot forging

摘要： Machine vision has been already used in measuring the dimensions of large hot forgings. However, the features of forgings are hard to extract robustness and efficiency, because of poor image quality caused by strong self-emitted radiation of hot forging. Therefore, in order to obtain clear images of large hot forgings, a-spectrum selection method based on SNR is proposed in this paper. First, the SNR model of interest features in images is established according to theory of image formation. Then, the SNRs of forging image in different temperature are estimated, and the wavelength range for a certain SNR is discussed, which can provide a necessary reference for capturing clear images of large hot forgings. A short-pass filter of a cut-off wavelength is selected to conduct experiments in forging workshop. Experimental results indicate that the spectrum selection method based on SNR is effective for measuring the dimensions of large hot forgings with machine vision technology. (C) 2015 Elsevier GmbH. All rights reserved.