Page 230

REVISTA IEEE 11

230 Journal of the Spanish Institute for Strategic Studies Núm. 11 / 2018 http://revista.ieee.es Photogrammetric techniques Photogrammetry allows us to determine the geometric properties of different objects from photographic images. Using photogrammetric techniques it is possible to obtain 3D information for the object space based on 2D information from the image space. For this purpose, the images must have been previously positioned, focused and rectified. Depending on the distance covered, photogrammetry can be either (a) close-up or (b) distant, the latter referring mainly to aerial photography. The process of reconstructing 3D objects through photogrammetry is based on the generation of three-dimensional clouds of points from pairs of stereo images (multi-view stereo). To determine the three-dimensional coordinates of any point in space it is necessary to have a set of images for the same point captured from different positions. On a larger scale, image-matching techniques are applied in order to locate common points in different images. The set-up of the images is crucial in terms of obtaining quality results. For this reason, the places from which the images are taken must be previously evaluated and analysed. A SIFT operator is generally used for finding homologous characteristics in the successive images37, thus facilitating the identification of the orientation and specific positions from which the successive images were taken. Subsequently, block adjustment of all the images is performed, a process whereby it is possible to determine a low-density cloud of 3D points. Within this process, a RANSAC robustness estimator is usually used to filter errors38. Although this point cloud gives us a fair idea of the shape of the scanned object, its density is too low for a precise reconstruction. For this reason, and in a second step, image-matching techniques are again applied to obtain a much denser point cloud, in which the main details of the scanned object can be more clearly distinguished. The SURE software, developed by the IFP, is used to densify point clouds from photogrammetric pairs39. The input required by the software is a set of oriented images, which are rectified as a first step. Then as a second step, SURE selects the most appropriate pairs of images to establish a matching process in a similar way as the Semi-Global Matching (SGM) algorithm does. Finally, the result obtained is a dense point cloud (in LAS format or similar), with resolution levels of one 3D point 37  LINGUA, Andrea; MARENCHINO, Davide and NEX, Francesco. «Performance analysis of the SIFT operator for automatic feature extraction and matching in photogrammetric applications», Sensors, 2009, n.º 9(5), pp. 3745-3766. 38  HAN, Lina; CHONG, Yanwen; LI, Yuanting and FRITSCH, Dieter. «3D Reconstruction by combining terrestrial laser scanner data and photogrammetric images». In: Proceedings of the Asia Assoc. of Remote Sensing, 2014, Nay Pyi Taw, Myanmar. 39  WENZEL, Konrad; ROTHERMEL, Mathias; HAALA, Norbert and FRITSCH, Dieter. «SURE – The IFP software for dense image matching». Photogrammetric Week ‘13, Ed. D. Fritsch, Wichmann: Berlin/Offenbach (Germany), pp. 59-70.


REVISTA IEEE 11
To see the actual publication please follow the link above