3D Stereo Mapping by Means of UltracamD Data
Publikation aus Digital
Posterpräsentation, Workshop 3D Remote Sensing in Forestry, 14.-15.2.2006 in Wien , 2006
The integration of the 3D component in the extraction of forest information from optical stereo data is a very recent research topic. Investigations carried out in the past have shown that it was already possible to derive information like mean stand height by mean of photogrammetric methods. However, it was found out that the possibilities of the derivation of very high resolution and highly detailed digital surface models (DSMs) from classical stereo data are limited when it comes to the modelling of detailed crown surface, single trees or small gaps in the forest. One alternative system is LiDAR, able to deliver highly detailed DSMs, but at quite high costs. Series of multi-overlapping, digital and very high resolution images can be seen as a high-qualitative as well as cost-effective alternative so far. In this study, images of UltracamD large format digital camera with an along-track overlap of 90% were used for the generation of such a detailed DSM. Due to the large overlap, new methods concerning stereo analysis are to be developed as the same object is visible from different viewing angles in several images. Based on the different viewing angles, valid results can be derived also for the problematic areas mentioned above. A “feature vector” matching method was used for image matching (Caballo-Perucha, 2003). Due to the multiple images involved, more than one matching result is obtained. For each point, the results with the highest quality are then used for the DSM generation. The sogenerated DSM is compared to a standard DSM (60% overlap, only two images) as well as to a Li-DAR DSM. The quality of the DSMs was assessed using in situ measured tree heights. It is shown, that the improvements of using more than two images are significant and the quality of the resulting DSM is almost comparable to the quality of the Lidar DSM.
Keywords: UltracamD, DSM generation, high resolution image data, digital airborne camera