Résumé - CAFD


Methods for improving the quality of a true orthomosaic of Vexcel UltraCam images created using a lidar digital surface model.

Benoît St-Onge.

The combined analysis of lidar and image datasets for information extraction of forest structural attributes and composition requires that the image-to-lidar geometric correspondence be known accurately. We propose a series of methods for producing a 10 cm high quality true orthomosaic of Vexcel UltraCam images perfectly adjusted to the lidar digital surface model (DSM). First, we introduce a technique for filling the small cavities visible on lidar raster DSMs. We then assess the image-to-lidar registration using visualization and quantitative approaches. The small geometric discrepancy measured between the two datasets is then corrected. In the image overlap areas, the true orthomosaic is created by choosing the contributing image that has the smallest distance to the corresponding DSM pixel. Occluded pixels that can not be seen from any centre of perspective are then filled with synthetic values calculated according to their sunlit/shadowed state at the time the images were taken. The resulting true orthomosaic is perfectly registered to the lidar dataset, is complete (considering occluded pixels receive synthetic values), is not radiometrically altered, and shows no visible cut lines. The proposed process should greatly help the simultaneous analysis of lidar and image datasets.