A Concept for Adaptive Mono-plotting Using Images and Laserscanner Data

The combination of photogrammetry (with its high geometric and radiometric resolution) and terrestrial laser scanning (allowing direct 3D measurement) is a very promising technique for object reconstruction, and has been applied for some time now, e.g. in the system Riegl LMS Z-420i. Nevertheless, the results presented from the combined laser-image-data very often are only coloured point clouds or textured meshes. Both object representations usually have erroneous representations of edges and corners (due to the characteristics of the laser measurement) and furthermore the amount of data to be handled in these “models” is typically enormous. In contrast to these object representations a surface model using a polyhedral compound would use only the relevant object points. However, the extraction of these modelling points from laser-image-data has not yet been fully automated. Especially the necessary generalization can only be accomplished by a human operator. Therefore, our aim is to support the operator in his work by speeding up the measurement of these modelling points. For this aim, this article presents a simple mono-plotting method that allows the human operator to identify each modelling point (on corners and edges) in the high-resolution images by a single mouse click. Subsequently, for this selected image ray, the missing distance is automatically determined from the associated laser data. This procedure starts by extracting the laser points in a cone around the image ray. Then these extracted points are tested for locally smooth surface patches (e.g. planar regions). Finally, the image ray is intersected with the foremost or hindmost of the extracted plane surface patches. Within this procedure the influence of erroneous laser measurements close to edges and corners can be avoided and furthermore, the distance from the scanner centre to the intersection point is determined with a better accuracy than the single laser point.