The data obtained during the measurement of optical 3D topography will directly affect the final measurement result. It is aimed at the noise and invalid measurement points that are difficult to identify and remove due to the hardware assembly error of the measurement system and the specular reflection during the structured light measurement. This article proposes a noise recognition and removal method based on the intermediate parameters of image processing. This method can make the measurement of point cloud data more accurate and reliable, and also avoid the more complicated and time-consuming point cloud data processing process. At the end of this article, experiments on noise identification and removal of large-scale point cloud data obtained by line laser scanning of engine blades were carried out. Experiments prove that this method can remove noise and invalid points in the point cloud under the premise of ensuring a certain accuracy, especially the measurement accuracy of the leaf edge.
Software And Hardware