Abstract
Orthorectification is an important step in generating accurate land use/land cover (LULC) from satellite imagery, particularly in urban areas with high-rise buildings. Such buildings generally appear as oblique shapes on very-high-resolution (VHR) satellite images, which reflect a bigger area of coverage than the real built-up area on LULC mapping. This drawback can cause not only uncertainties in urban mapping and LULC classification, but can also result in inaccurate urban change detection. Overestimating volume or area of high-rise buildings has a negative impact on computing the exact amount of environmental heat and emission. Hence, in this study, we propose a method of orthorectfiying VHR WorldView-3 images by integrating light detection and ranging (LiDAR) data to overcome the aforementioned problems. A 3D rational polynomial coefficient (RPC) model was proposed with respect to high-accuracy ground control points collected from the LiDAR data derived from the digital surface model. Multiple probabilities for generating an orthrorectified image from WV-3 were assessed using 3D RCP model to achieve the optimal combination technique, with low vertical and horizontal errors. Ground control point (GCPs) collection is sensitive to variation in number and data collection pattern. These steps are important in orthorectification because they can cause the morbidity of a standard equation, thereby interrupting the stability of 3D RCP model by reducing the accuracy of the orthorectified image. Hence, we assessed the maximum possible scenarios of resampling and ground control point collection techniques to bridge the gap. Results show that the 3D RCP model accurately orthorectifies the VHR satellite image if 20 to 100 GCPs were collected by convenience pattern. In addition, cubic conventional resampling algorithm improved the precision and smoothness of the orthorectified image. According to the root mean square error, the proposed combination technique enhanced the vertical and horizontal accuracies of the geo-positioning process to up to 0.8 and 1.8 m, respectively. Such accuracy is considered very high in orthorectification. The proposed technique is easy to use and can be replicated for other VHR satellite and aerial photos.
Highlights
The rapid expansion of remote sensing technologies has greatly upgraded the spectral and spatial resolutions of remotely sensed images [1,2]
Vertical and horizontal accuracies were assessed via statistical analysis in the following variation sources: (i) pattern of ground control points (GCPs) collection from light detection and ranging (LiDAR) digital surface models (DSMs), (ii) number of GCPs used in the triangulation process and (iii) resampling algorithms of in the orthorectification process
The linear error (LE) metric evaluates the difference between the GCP-measured elevation and the LiDAR DSM elevation with an optional geoid offset in meters at the 95% confidence level using
Summary
The rapid expansion of remote sensing technologies has greatly upgraded the spectral and spatial resolutions of remotely sensed images [1,2]. Numerous mathematical models for VHR satellite sensor orientation and geo-positioning using GCPs have been investigated to correct geometric distortions from imagery [20,21] Those models can be generally classified as physical and deterministic or empirical, whereas each class can be represented in 2D or 3D method [20]. Comprehensive topographical maps (scale 1:5000 or finer) and remotely sensed imagery with the similar spatial resolution of the PAN band of an orthorectified satellite image, such as light detection and ranging (LiDAR) dataset, are appropriate for extracting horizontal and vertical coordinates for GCPs. The 2D Polynomial functions are not perfect systems to orthorectify VHR images because these imageries did not consider the effects of ground elevation which must be corrected for basic planimetric distortion at the GCPs [30,31]. Vertical and horizontal accuracies were assessed via statistical analysis in the following variation sources: (i) pattern of GCP collection from LiDAR DSM, (ii) number of GCPs used in the triangulation process and (iii) resampling algorithms of in the orthorectification process
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
