An improved image acquiring method for machine vision measurement of hot formed parts

Non-contact dimensional measurements of forgings are desirable to permit real-time process control. Machine vision technology has been widely employed as a high-speed, non-contact method in dimensional measurement. However, the extracted features of high-temperature forgings are not suitable for application because of self-emitted radiation. This paper primarily introduces a spectrum selective method for improving the image quality of hot parts. By filtering certain wavelength range light and projecting light to hot parts surface, the images of the hot parts can be obtained without interference from self-emitted radiation. Experiments were also conducted to indicate that the images of hot parts were greatly improved and the feature points could be extracted and matched using the proposed method.