Dimensional Measurement of Small Hot Pieces Based on a Monochrome CCD

Abstract

For the purpose of controlling the quality of forging process, a low-priced monochrome CCD is adopted to establish a vision measurement system to gauge the two-dimensional sizes of hot parts accurately in the paper. To derive the two dimensions of the forgings, comparison measurement method is employed. In the system, images are taken by the CCD through an infrared filter to filter out the strong light emitted from the forgings. And then average filtering and morphological filtering are used to reduce the effect of image noise. In order to improve the precision of geometry position, image distortion is corrected to reduce the optical distortion errors. After that, the edges in images are extracted at sub-pixel level to improve the location accuracy of geometry features. After calibration, the horizontal equivalent size of a pixel is 0.119mm, while the vertical 0.120mm. As be verified, the horizontal measurement uncertainty is 0.0069mm, while vertical 0.0063mm. Images of a 45 # steel cuboid at 1000 are captured and its dimensions are calculated through the procedure described in the paper, which have a difference less than 0.5mm compared with the theoretical values, indicating the precision and practicality of the method proposed in the paper.

Metrological Characterization of a Laser-Camera 3d Vision System Through Perspective-N-Point Pose Computation and Monte Carlo Simulations

“…Profilometers typically use blue or green lasers and triangulation in order to reconstruct a part of the section of the workpiece. These sensors allow obtaining uncertainties in the order of mm when measuring diameters or linear dimensions of approximately 1 m. Standard cameras can be used to detect the diameter or the silhouette of the objects, with accuracies that are 1 order of magnitude larger than those of the profilometers; however, camera allows retrieving dense information about the specimen, allowing to reconstruct the entire geometry of the workpiece during its rotation (Chao et al 2015;Ghiotti et al 2015;Nye, Elbadan, and Bone 2001;Zhang et al 2014). The main idea behind this work is to use data fusion in order to combine the density of information of the camera-based measurements with the accuracy of 2D profilometers.…”