Entradas

Accuracy of Digital Surface Models and Orthophotos Derived from Unmanned Aerial Vehicle Photogrammetry

/0 Comentarios/en , /por

Francisco Agüera Vega, Fernando Carvajal Ramírez, Patricio J. Martínez Carricondo

2016

Journal of Surveying Engineering, 04016025

http://dx.doi.org/10.1061/(ASCE)SU.1943-5428.0000206

ABSTRACT

This paper explores the influence of flight altitude, terrain morphology, and the number of ground control points (GCPs) on digital surface model (DSM) and orthoimage accuracies obtained with unmanned aerial vehicle (UAV) photogrammetry. For this study, 60 photogrammetric projects were carried out considering five terrain morphologies, four flight altitudes (i.e., 50, 80, 100, and 120 m), and three different numbers of GCPs (i.e., 3, 5, and 10). The UAV was a rotatory wing platform with eight motors, and the sensor was a nonmetric mirrorless reflex camera. The root-mean-square error (RMSE) was used to assess the accuracy of the DSM (Z component) and orthophotos (X, Y, and XY components RMSEX, RMSEY, and RMSEXY, respectively). The results show that RMSEX, RMSEY, and RMSEXY were not influenced by flight altitude or terrain morphology. For horizontal accuracy, differences between terrain morphologies were observed only when 5 or 10 GCPs were used, which were the best accuracies for the flattest morphologies. Nevertheless, the number of GCPs influenced the horizontal accuracy; as the number of GCPs increased, the accuracy improved. Vertical accuracy was not influenced by terrain morphology, but both flight altitude and the number of GCPs had significant influences on RMSEZ; as the number of GCPs increased, the accuracy improved. Regarding flight altitude, vertical accuracy decreased as flight altitude increased. The most accurate combination of flight altitude and number of GCPs was 50 m and 10 GCPs, respectively, which yielded RMSEX, RMSEY, RMSEXY, and RMSEZ values equal to 0.038, 0.035, 0.053, and 0.049 m, respectively.
In view of these results, the map scale according to the legacy American Society for Photogrammetry and Remote Sensing map standard of 1990 will be approximately 1:150, and an equivalent contour interval of 0.150 m is sufficient for most civil engineering projects.

Geometric accuracy of Ikonos Geo Panchromatic orthoimage products

/0 Comentarios/en , /por

Manuel A. Aguilar, Fernando J. Aguilar, Fernando Carvajal Ramírez, Francisco Agüera Vega,

2006

Revue Française de Photogrammétrie et de Telédétection. ISPRS Commission Technique I, Symposium, nº 184: 5-10

http://www.scopus.com/inward/record.url?eid=2-s2.0-33748913533&partnerID=MN8TOARS

ABSTRACT

The new very high space resolution satellite images, such as QuickBird and IKONOS, open new possibilities in cartographic applications. This work has as its main aim the assessment of different sensor models for achieving the best geometric accuracy in orthorectified imagery products obtained from IKONOS Geo Ortho Kit Imagery. Two dimensional Root Mean Square Error (RMSE2D) is computed and utilized as accuracy indicator. The ancillary data were generated by high accuracy methods: (1) Check (ICPs) and control points (GCPs) were measured with a differential global positioning system (DGPS) and, (2) a digital elevation model (DEM) with grid spacing of 5 m derived from digitized contour lines with an interval of 10 m and extracted from the 1:10,000 Andalusia Topographic Maps series (RMSEz<1.75 m), was used for image orthorectification process. Four sensor models were used to correct the satellite data: (1) First order 3D rational functions without vendor image support data (RFM1), (2) 3D rational functions refined by the user with zero order polynomial adjustment (RPC0), (3) 3D rational functions refined by the user with first order polynomial adjustment (RPC1), and (4) the 3D Toutin physical model (CCRS). The number of control points per orthorectified imagery (9 and 18 GCPs) and their distribution (random and stratified random sampling) were studied as well. The best results, both in the phase of sensor orientation (RMSEO about 0.59 m) as in the final orthoimages (RMSEORTHO about 1.25 m), were obtained when the model RPC0 was used. Neither a large number of GCPs (more than nine) nor a better distribution (stratified random sampling) improved the results obtained from RPC0.