PE&RS July 2005

Peer-Reviewed Articles

805 Effects of Terrain Morphology, Sampling Density, and Interpolation Methods on Grid DEM Accuracy
Fernando J. Aguilar, Francisco Agüera, Manuel A. Aguilar, and Fernando Carvajal

Abstract  Download Full Article
This paper explores the effects of terrain morphology, sampling density, and interpolation methods for scattered sample data on the accuracy of interpolated heights in grid Digital Elevation Models (DEM). Sampled data were collected with a 2 by 2 meters sampling interval from seven different morphologies, applying digital photogrammetric methods to large scale aerial stereo imagery (1:5000). The experimental design was outlined using a factorial scheme, and an analysis of variance was carried out. This analysis yielded the following main conclusions: DEM accuracy (RMSE) is affected significantly by the variables studied in this paper according to “morphology > sampling density > interpolation” method. Multiquadric Radial Basis Function (RBF) was rated as the best interpolation method, although Multilog RBF performed similarly for most morphologies. The rest of RBF interpolants tested (Natural Cubic Splines, Inverse Multiquadric, and Thin Plate Splines) showed numerical instability working with low smoothing factors. Inverse Distance Weighted interpolant performed worse than RBF Multiquadric or RBF Multilog. In addition, it is found that the relationship between the RMSE and the sampling density N is adjusted to a decreasing potential function that may be expressed as RMSE/Sdz = 0.1906(N/M)^-0.5684 (R² = 0.8578), being Sdz the standard deviation of the heights of the M check points used for accuracy estimation, and N the number of sampling points used for creating the DEM. The results obtained in this study allow us to observe the possibility of establishing empirical relationships between the RMSE expected in the interpolation of a Grid DEM and such variables as terrain ruggedness, sampling density, and the interpolation method, among others that could be added. Therefore, it would be possible to establish a priori the optimum grid size required to generate or storage a DEM of a particular accuracy, with the economy in computing time and file size that this would signify for the digital flow of the mapping information.

817 An Evaluation of Lidar-derived Elevation and Terrain Slope in Leaf-off Conditions
Michael E. Hodgson, John Jensen, George Raber, Jason Tullis, Bruce A. Davis, Gary Thompson, and Karen Schuckman

Abstract  Download Full Article
The effects of land cover and surface slope on lidar-derived elevation data were examined for a watershed in the piedmont of North Carolina. Lidar data were collected over the study area in a winter (leaf-off) overflight. Survey-grade elevation points (1,225) for six different land cover classes were used as reference points. Root mean squared error (RMSE) for land cover classes ranged from 14.5 cm to 36.1 cm. Land cover with taller canopy vegetation exhibited the largest errors. The largest mean error (36.1 cm RMSE) was in the scrub-shrub cover class. Over the small slope range (0° to 10°) in this study area, there was little evidence for an increase in elevation error with increased slopes. However, for low grass land cover, elevation errors do increase in a consistent manner with increasing slope. Slope errors increased with increasing surface slope, under-predicting true slope on surface slopes >2°. On average, the lidar- derived elevation under-predicted true elevation regardless of land cover category. The under-prediction was significant, and ranged up to -23.6 cm under pine land cover.

825 Structural Damage Assessments from Ikonos Data Using Change Detection, Object-Oriented Segmentation, and Classification Techniques
D.H.A. Al-Khudhairy, I. Caravaggi, and S. Giada

Abstract  Download Full Article
Recent improvements in the spatial resolution of commercial satellite imagery make it possible to apply very high-resolution (VHR) satellite data for assessing structural damage in the aftermath of humanitarian crises, such as, armed conflicts. Visual interpretation of pre- and post-crisis very high-resolution satellite imagery is the most straightforward method for discriminating structural damage and assessing its extent. However, the feasibility of using visual interpretation alone diminishes in the cases of large and dense urban settlements and spatial resolutions in the range of 2 m to 3 meters and larger. Visual interpretation can be further complicated at spatial resolutions greater than 1 m if accompanied by shadow formation and differences in sensor and solar conditions between the pre- and post-conflict images.

In this study, we address these problems through investigating the use of traditional change techniques, namely, image differencing and principle component analysis, with an object-oriented image classification software, e-Cognition. Pre-conflict Ikonos (2 m resolution) images of Jenin in the Palestinian territories and Brest (1 m resolution) in FYROM were classified using the e-Cognition software. Thereafter, the pre-conflict classification was used to guide the classification, using e-Cognition, of the pixel-based change detection analysis. The second part of the study examines the feasibility of using mathematical morphological operators to automatically identify likely structurally damaged zones in dense urban settings. The overall results are promising and show that object-oriented segmentation and classification systems facilitate the interpretation of change detection results derived from very high-resolution (1 m and 2 m) commercial satellite data. The results show that object-oriented classification techniques enhance quantitative analysis of traditional pixel-based change detection applied to very high-resolution satellite data and facilitate the interpretation of changes in urban features. Finally, the results suggest that mathematical morphological methods are a potential new avenue for automatically extracting likely damaged zones from very high-resolution satellite imagery in the aftermath of disasters.

839 Sub-pixel Target Mapping from Soft-classified, Remotely Sensed Imagery
Peter M. Atkinson

Abstract  Download Full Article
A simple, efficient algorithm is presented for sub-pixel target mapping from remotely-sensed images. Following an initial random allocation of “soft” pixel proportions to “hard” sub-pixel binary classes, the algorithm works in a series of iterations, each of which contains three stages. For each pixel, for all sub-pixel locations, a distance-weighted function of neighboring sub-pixels is computed. Then, for each pixel, the sub-pixel representing the target class with the minimum value of the function, and the sub-pixel representing the background with the maximum value of the function are found. Third, these two sub-pixels are swapped if the swap results in an increase in spatial correlation between sub-pixels. The new algorithm predicted accurately when applied to simple simulated and real images. It represents an accessible tool that can be coded and applied readily by remote sensing investigators.

847 DTM Generation and Building Detection from Lidar Data
Ruijin Ma and Will Meyer

Abstract  Download Full Article
Object reconstruction has attracted great attention from both computer vision and photogrammetry communities, and new technologies are being introduced into this research society. Lidar (Light Detection And Ranging) has become well recognized in the geomatics community since the late 1990s. Compared with traditional photogrammetry, lidar has advantages in measuring surface in terms of accuracy and density, automation, and fast delivery time. There is a large market in geo-data acquisition and object recognition for lidar technology (Baltsavias, 1999). In a general sense, lidar is a companion technology for traditional photogrammetry. The direct product that can be derived from lidar data is the DSM (Digital Surface Model), which depicts the topography of the earth’s surface, including objects above the terrain. Further processing can be carried out to generate DEM (Digital Terrain Model) and object models like buildings, which is very useful information in telecommunication, city planning, flood control, and tourism. Morphology and classification are two commonly used methods in DEM generation and object reconstruction. However, these two methods are either sensitive to errors or of low accuracy. In this paper, a new method is proposed to extract ground points for DEM generation and to detect points belonging to buildings. A new method for boundary regularization is also proposed. The results show that buildings can be detected with high accuracy from lidar data.

855 A Split-and-Merge Technique for Automated Reconstruction of Roof Planes
Kourosh Khoshelham and Zhilin Li

Abstract  Download Full Article
Automated reconstruction of buildings from different data sources has been one of the most challenging problems in photogrammetry and computer vision. Systems for automated building reconstruction fail in many cases due to complexities involved in the data including image noise, occlusion, shadow, and low contrast, as well as, low accuracy or density of height data. In this paper, the problem of overgrown and undergrown regions in the segmentation of aerial images is discussed, and a split-and-merge technique is presented to overcome this problem by making use of height data. This technique is based on splitting image regions whose associated height points do not fall in a single plane, and merging coplanar neighboring regions. A robust plane-fitting method is used to fit planar surfaces to height points that are highly contaminated by gross errors. Final roof planes are extracted out of the image planar regions by checking their slope and height over a morphologically opened DSM. An experimental evaluation is conducted, and its results indicate the capability of the proposed technique in splitting overgrown regions, merging undergrown coplanar regions, and selecting the final roof planes. Also, the method is shown to be computationally efficient, and the reconstructed roof planes are of acceptable accuracy.

863 Theoretical Analysis of the Iterative Photogrammetric Method to Determining Ground Coordinates from Photo Coordinates and a DEM
Yongwei Sheng

Abstract  Download Full Article
It is necessary to determine ground coordinates from a single aerial image and a digital elevation model (DEM). A widely used method in photogrammetry is to iteratively calculate the coordinates based on the inverse collinearity equations. The iterative photogrammetric method may be divergent when the terrain surface becomes complicated. However, there is a lack of theoretical analysis of this method. This paper theoretically analyzes the convergence condition and the convergence speed of the method, and validates the theory using a simulated surface containing various slope conditions. The elevation angleθ of the view ray and the inclination angle of the profile, intersected by the terrain surface and the vertical view plane, play a critical role in the method. The necessary and sufficient condition for convergence is < . The convergence speed is quantified as a function of , θ, the convergence threshold T and the offset Z₀ of initial elevation estimate.