In order to improve the quality and efficiency of forging process, it needs to execute on-line dimensional measurement of the forgings. In the paper, a laboratory color vision measuring system is set up and the combination of digital and physical filtering is adopted to improve the image quality based on the radiation characteristics of high-temperature forgings. The digital filtering technology is a kind of image processing methods, in which the R component of the forging image is removed. While, the physical filtering technology is achieved by optical filters installed in front of the CCD, in which strong self-emitted radiation from the hot parts can be filtered out. In order to evaluate the image quality, the image contrast is applied, which is generally defined as the difference value between average gray scale of object region and that of background region. In the experiments, image contrast derived with filters at different sample points set from 800°C to 1200°C is compared to determine the optimal scheme of filters to be selected. Results of experiments indicate that the application effect of filters is dissimilar when the forging is in different temperature ranges. Through comparison, the optimal selection scheme of filters is determined to derive high quality image of forgings at different temperatures, which lays a solid foundation for the subsequent image processing.