Geometry measurements with image processing and forward projection

The automatic inspection of 2D or 3D geometries are used e.g. for component control in production processes. Often, such a visual inspection uses cameras and image processing to measure geometries. Geometry measurement methods use e.g. triangulation or stereo cameras. We propose a new approach based on our presented camera calibration using re-projection as well as forward projection (e.g. DGaO Proceedings 2022). After calibration, the 2D image coordinates of a 3D object are forward projected into the 3D world. Only once the distance and one dimension of a 3D object needs to be known by our method. For different distances, the scaling value s from our forward projection algorithm indicates the new distance and thus we measure the geometry of the object. Even if the object is translated or rotated, the forward projection can predict the geometry of the object. We recognize different standard flat geometries using image processing. Then, we apply forward projection for different distances using our method to predict the geometric sizes of the object. We will present our results obtained from image processing and forward projection with real measurements and discuss precision.