DSM Reconstruction from Uncalibrated Multi-View Satellite Stereo Images by RPC Estimation and Integration

This research is concerned with the use of remotely sensed data in the estimation of the solar energy potentiality of the flat roofs in El Maadi District, Cairo. A solar radiation model was applied using a digital surface model (DSM) generated from two Worldview stereo satellite images using the digital photogrammetric workstation (Leica Photogrammetric Suite) LPS. The procedures included applying orientation with Rational Polynomial Coefficients (RPC) with and without GCPs, tie point measurements, aerial triangulation, automatic DSM and editing, true ortho-rectification. The accuracy of the DSM has been assessed and reported for both methods. The ortho-rectification, of stereo satellite images has been performed. A PAN true ortho-images with 0.5 m resolution resulted from applying orientation with rational polynomial coefficients (RPCs) refined with four accurate differential ground control points (DGCPs) and a high accurate DSM derived from stereo images were generated. Two classification methods were conducted (SVM and ANN) on the true ortho-image. First method without utilizing texture, the second utilizing texture features of Grey Level Co-occurrence Matrix (GLCM) as inputs. The SVM of the ortho-image and texture features overall accuracy was 93.4%, kappa coefficient was 0.92 and ANN of ortho-image and texture features overall accuracy of 91.2% and kappa coefficient was 0.90. Using the generated DSM as input, the area solar radiation model in ESRI ArcGIS was run. The model resulted in an annual total radiation grid for year 2015 which was converted to average daily radiation in kilowatt hour/square meters per day. A sample of 265 buildings footprints was extracted from the classification and used for the estimation. Assuming that 50% roof areas are available for solar Photovoltaic (PV), 12% PV grid conversion efficiency and 0.6 performance ratio and using ESRI zonal statistics tool, the technical potentiality for solar PV electricity generation for the sample buildings was calculated, classified into five classes and mapped.

In this paper, we analysed the geometric model characteristics based on rational polynomial coefficient (RPC) model using high resolution Pleiades satellite stereo images i.e., Pleiades 1A and 1B satellite sensors. 3D maps were generated for objects like trees near poles of power transmission lines using satellite stereo images. This study was very challenging due to weather condition, hilly and non-uniform terrain at Kota kina balu in East Malaysia. First we require the geometric accuracy of these two sensors for 3D mapping of the area of interest. Also we require accurate Digital elevation model for better height calculation of vegetation/trees near power poles. For these purposes, we evaluated the geometric accuracy of both Pleiades sensors i.e., 1A and 1B which is based on direct rational function model and inverse Rational Function model (RFM). The comparison was made based on RPC coefficients. Results shown that values of RPC coefficients for Pleiades 1A sensor are less as compared to the Pleiades 1B sensor. This shows that Pleiades 1A sensor can be used for monitoring the vegetation near poles of high transmission power lines based on the satellite stereo images.

ZiYuan-3 (ZY-3, meaning Resource-3) is China's first civilian multispectral satellite with high-resolution stereo mapping capability. The along-track stereo images of ZY-3 allow the generation of digital elevation models (DEMs) for various mapping applications. This article evaluated the precision of the derived stereo DEM and orthoimages of ZY-3 data for a test site in Northeastern China. Pixel offset and GPS survey data were used to evaluate the image products. Orthoimages were derived by incorporating rational polynomial coefficients (RPCs) of the satellite orbits and the reference DEMs, including the Advanced Spaceborne Thermal Emission and Reflection Radiomete (ASTER) global digital elevation model (GDEM) and ZY-3 stereo elevation model. The generations of orthoimages were categorised into four approaches depending on the RPCs, control points and/or reference DEMs used. When only the ASTER GDEM was used, the derived orthoimages of the forward and backward viewing sensors showed clear pixel offsets against the GPS check points (CKPs). For example, the offsets of the orthoimage derived using forward-looking image were 13·58 and 26·50 m, in x (easting) and y (northing) directions, respectively. These results also indicated that the pixel offset vectors between the forward and backward orthoimages was systematic which was caused by the error in the RPCs. The second and third approaches were to generate the ZY-3 stereo DEMs first, with and without the use of ground control points (GCPs), and then used the stereo DEMs to generate the orthoimages of the forward and backward looking images. The forth approach used both GCPs and ASTER GDEM to generate the orthoimages. The results showed the use of four GCPs in the process improved the accuracy of the orthoimages. Therefore, ZY-3 can deliver the DEM and orthoimage products to meet the accuracy requirement of the national 1 : 50 000 mapping with four well distributed GCPs.

Thanks to rational polynomial coefficients (RPCs), which are provided by vendors to end users, digital elevation models (DEMs) can be simply derived from satellite stereo images. However, DEMs are influenced by systematic errors in the rational function model (RFM), known as RPC biases. Global DEMs (GDEMs), such as the Shuttle Radar Topography Mission (SRTM), which is the most inexpensive solution, can be applied to improve the accuracy of the relative RFM-derived DEMs. In this article, an automatic and robust local feature-based DEM matching and orientation approach is proposed in order to improve the accuracy of the relative RFM-derived DEMs without the use of ground control points (GCPs). The proposed approach consists of four main steps: (1) combined local feature extraction; (2) computation of the distinctive order-based self-similarity (DOBSS) descriptor; (3) a feature correspondence and local consistency checking process; and (4) a relative RFM-derived DEM orientation process using three-dimensional (3D) transformation models, including 3D rigid, 3D similarity and 3D affine transformations. This technique can avoid the sensitivity of conventional 3D DEM matching methods to initial values, monotonous areas and local distortions. Experimental results on two CARTOSAT-1 derived DEMs demonstrate the superior performance of the proposed DEM matching method over state-of-the-art methods, including SIFT, DAISY, LIOP, LBP, and BRISK descriptors, in terms of the number of correct matches (NCM) and DEM orientation accuracy. The results also show that the proposed method is able to significantly improve the geometric accuracy of the relative RFM-derived DEMs.

Generation of precise digital elevation models (DEMs) from stereo satellite images by using rational polynomial coefficients (RPCs) usually needs several ground control points (GCPs). This is mainly due to RPCs biases. However, since GCPs collection is a time consuming and expensive process, global DEMs (GDEMs), as the most inexpensive geospatial information, can be used to improve stereo satellite imagery-based DEMs (IB-DEMs). In this study, a 2.5 D mutual information based DEM matching, between a GDEM and an IB-DEM, was introduced for bias correction of satellite stereo images. Three well-known 30-meter GDEMs, namely, SRTM, ASTER, and AW3D30, were used and compared to assess the efficiency of this approach. The performance of the proposed method was evaluated by processing the stereo images acquired by CARTOSAT-1 satellite from two regions with flat, hilly, and mountainous topography. Evaluation results revealed that the proposed method could significantly improve the geometric accuracy of IB-DEM using all GDEMs.

ABSTRACTThe overfitting phenomenon and rational polynomial coefficients (RPCs) biases are two crucial issues that degrade the accuracy of geospatial products derived from high-resolution satellite images. The overfitting phenomenon is caused by both a large number of RPCs and strong correlations among them. The RPC biases arise from uncertainties in the global positioning system receivers and inertial measurement units. In this article, an innovative framework based on the genetic algorithm (GA) and the least squares (LS) algorithm, called GALS, is proposed to overcome these problems simultaneously. In this method, the GA is applied to select the optimum RPCs, while the LS algorithm is used to estimate the values of the optimally selected RPCs. The GALS method requires various sets of well-distributed ground control points (GCPs). To tackle the problem of GCP collection, we generated a large number of digital elevation model (DEM)-derived GCPs (DEMGCPs), using a global DEM (GDEM) and vendor-provided RPCs, refined by only one GCP. To evaluate the performance of this framework, four IRS-P5 data sets were used. The GALS is compared to two competing methods, L1-norm-regularized LS and ridge estimation by considering two scenarios using 50 GCPs and the DEMGCPs. The results demonstrate the superiority of GALS in both scenarios. Furthermore, GALS using DEMGCPs led to far more accurate and stable results when compared to GALS using GCPs. Compared to the vendor-provided RPCs, the results of the GALS using DEMGCPs also indicate a major improvement, single-pixel or subpixel accuracy with around 15 RPCs, and only 1 GCP, in both accuracy and reliability of georeferencing for all IRS-P5 data sets.

KOMPSAT-2와 같은 고해상 위성영상은 대상영역의 3차원 위치결정을 위하여 RPC(Rational Polynomial Coefficient)가 포함된 자료를 제공한다. 그러나 RPC로 계산된 영상기하는 일정량의 편이(systematic errors)를 지니고 있는 상태이며, 이를 보정하기 위해서는 수 개 이상의 지상기준점(ground control point)이 필요하다. 이에 본 논문에서는 지상기준점 없이 입체영상(stereo pair)과 SRTM(Shuttle Radar Topography Mission) DEM(Digital Elevation Model) 사이의 대응점(tie point)만을 이용하여 자동으로 영상 기하를 보정하는 효과적인 방법을 제안하였다. 이러한 방법은 4가지 단계를 포함 한다: 1) 대응점 추출, 2) 대응점에 대한 지상좌표 결정, 3) SRTM DEM을 이용한 지상좌표의 보정, 4) RPC 보정 모델의 파라미터 결정. 우리는 KOMPSAT-2 입체영상을 이용하여 제안된 방법의 성과를 입증하였다. 검사점(check point)을 통해 계산된 RMSE(Root Mean Square Error)는 X와 Y, Z방향으로 각각 약 3.55 m, 9.70 m, 3.58 m를 나타냈다. 이는 SRTM DEM을 이용하여 RPC가 지닌 편이를 X, Y 및 Z 모든방향에 대하여 10 m이내의 정확도로 자동보정할 수 있다는 것을 의미한다. A high resolution satellite imagery such as KOMPSAT-2 includes a material containing rational polynomial coefficient (RPC) for three-dimensional geopositioning. However, image geometries which are calculated from the RPC must have inevitable systematic errors. Thus, it is necessary to correct systematic errors of the RPC using several ground control points (GCPs). In this paper, we propose an efficient method for automatic correction of image geometries using tie points of a stereo pair and the Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM) without GCPs. This method includes four steps: 1) tie points extraction, 2) determination of the ground coordinates of the tie points, 3) refinement of the ground coordinates using SRTM DEM, and 4) RPC adjustment model parameter estimation. We validates the performance of the proposed method using KOMPSAT-2 stereo pair. The root mean square errors (RMSE) achieved from check points (CPs) were about 3.55 m, 9.70 m and 3.58 m in X, Y;and Z directions. This means that we can automatically correct the systematic error of RPC using SRTM DEM.

Abstract. High resolution, optical satellite sensors are boosted to a new era in the last few years, because satellite stereo images at half meter or even 30cm resolution are available. Nowadays, high resolution satellite image data have been commonly used for Digital Surface Model (DSM) generation and 3D reconstruction. It is common that the Rational Polynomial Coefficients (RPCs) provided by the vendors have rough precision and there is no ground control information available to refine the RPCs. Therefore, we present two relative orientation methods by using corresponding image points only: the first method will use quasi ground control information, which is generated from the corresponding points and rough RPCs, for the bias-compensation model; the second method will estimate the relative pointing errors on the matching image and remove this error by an affine model. Both methods do not need ground control information and are applied for the entire image. To get very dense point clouds, the Semi-Global Matching (SGM) method is an efficient tool. However, before accomplishing the matching process the epipolar constraints are required. In most conditions, satellite images have very large dimensions, contrary to the epipolar geometry generation and image resampling, which is usually carried out in small tiles. This paper also presents a modified piecewise epipolar resampling method for the entire image without tiling. The quality of the proposed relative orientation and epipolar resampling method are evaluated, and finally sub-pixel accuracy has been achieved in our work.

The purpose of this research was to analyze the quality and characteristics of four digital elevation models (DEMs) generated using in-track Korea Multi-Purpose Satellite (KOMPSAT)-3 stereo images. The sensor modeling methods were based on ground control points (GCPs), the initial rational polynomial coefficients (RPCs), relative adjustment, and the automatic bias-compensation method. The GCPs and check points (CPs) were extracted from the 0.25 m aerial orthoimage and the 5 m DEM provided by the National Geographic Information Institute (NGII). The DEMs had the same resolution as the reference DEM (5 m) and comparative analysis was carried out. The results indicate that when relative adjustment was applied alone (DEM 3), the percentage of matched points with a correlation of 0.8 or more was improved by at least 17% compared to the case where only initial RPCs were used (DEM 2). Although the absolute horizontal position error of DEM 3 could not be eliminated, the relative elevation error at the same position was reduced significantly. Therefore, if the relative positions of DEMs produced at different times can be corrected, they can be used for the detection of changes in altitude. When applying the automatic bias-compensation method (DEM 4) without GCPs, the percentage of matched points with a correlation of 0.8 or more was 70.1%. When GCPs were used (DEM 1), the value was 70.2%, i.e., almost identical to that of DEM 4. The mean difference in resolution among DEMs 1 and 4 was -1.8 ± 3.4 m (median, -1.0 m). The results show that DEMs of sufficient quality can be generated without GCPs. Furthermore, although discrepancies among the DEMs were noted in forest and shadow areas, it is possible to produce a 5~10 m resolution DEM by using additional image processing techniques, such as shadow removal.

Summary The following study proposes a novel method for creation of Digital elevation model (DEM) which can be used to monitor the changes that take place in glaciers and complete the glacier inventory. DEM was created using ASTER stereo images, with Gangotri Glacier being the study area. Data obtained from ASTER Satellite provides Rational Polynomial Coefficient (RPC) which facilitated us in the mapping from object space (longitude, latitude, height) to image space (sample, line). Ground Control Points (GCPs) are crucial to improve the geographic positions predicted by the supplied RPCs. Pan-sharpened Colour composite of bands 5-4-3 obtained from Landsat OLI satellite and SRTM (30 m resolution) DEM was used for collecting uniformly distributed GCPs. Good Image matching of both the stereo pairs was done using Tie Points. ASTER DEM was then generated, following the standard procedure in “ENVI 5.0.” The resultant elevation error, assessed by comparing the elevations derived from SRTM and ASTER was found to be around ±30m. This study will help, further in monitoring temporal Volumetric and Mass changes over the Gangotri region. Our methodology has the potential to quickly produce and verify the DEMs in high mountainous regions efficiently.

2012년 5월 발사된 다목적 실용위성 아리랑 3호는 단일 패스 상에서 0.7m의 공간해상도로 스테레오 영상을 획득할 수 있어 기존의 아리랑 2호에 비해 고품질의 Digital Elevation Model(DEM) 추출이 가능하다. 통상적으로 DEM 추출을 위해서는 영상 전반에 걸쳐 골고루 취득된 정밀한 기준점을 사용하여 센서모델링을 수행하고, 스테레오 매칭을 수행해야하나, 해외 지역이나 접근이 힘든 지역 등 GPS측량이 쉽지 않은 지역의 경우에는 무기준점 또는 기준점을 최소화하거나 기 구축된 공간 데이터를 활용하는 등의 방법으로 DEM을 추출해야 한다. 따라서 본 연구에서는 아리랑 3호 데이터로부터 무기준점 기반, 상대표정 기반, 단일 기준점 등 여러 가지 Rational Polynomial Coefficients(RPC) 처리 조건에 따라 DEM을 생성하고 정확도를 평가하였다. 기 구축된 공간영상인 Digital Orthophoto Quadrangle(DOQ) 와 Shuttle Radar Topography Mission(SRTM)을 기준점 자료로 활용하여 미국지역 아리랑3호 스테레오 데이터를 대상으로 DEM을 생성하였으며, LiDAR DEM을 이용하여 정확도를 평가하였다. 실험 결과 무기준점인 경우 상대표정을 통해 의미 있는 정확도 향상을 얻을 수 있었고, DOQ와 SRTM 조합의 단일 기준점으로도 영상 전반에 걸쳐 기준점을 획득한 경우에 근접하는 7m 정도의 DEM 정확도를 확보할 수 있었다. Kompsat-3 is an optical high-resolution earth observation satellite launched in May 2012. In addition to its 0.7m spatial resolution, Kompsat-3 is capable of in-track stereo acquisition enabling quality Digital Elevation Model(DEM) generation. Typical DEM generation procedure requires accurate control points well-distributed over the entire image region. But we often face difficult situations especially when the area of interests is oversea or inaccessible area. One solution to this is to use existing geospatial data even though they only cover a part of the image. This paper aimed to assess accuracy of DEM from Kompsat-3 with different scenarios including no control point, Rational Polynomial Coefficients(RPC) relative adjustment, and RPC adjustment with control points. Experiments were carried out for Kompsat-3 stereo data in USA. We used Digital Orthophoto Quadrangle(DOQ) and Shuttle Radar Topography Mission(SRTM) as control points sources. The generated DEMs are compared to a LiDAR DEM for accuracy assessment. The test results showed that the relative RPC adjustment significantly improved DEM accuracy without any control point. And comparable DEM could be derived from single control point from DOQ and SRTM, showing 7 meters of mean elevation error.

The correct representation of the topography of terrain is an important requirement to generate photogrammetric products such as orthoimages and maps from high-resolution (HR) or very high-resolution (VHR) satellite datasets. The refining of the digital elevation model (DEM) for the generation of an orthoimage is a vital step with a direct effect on the final accuracy achieved in the orthoimages. The refined DEM has potential applications in various domains of earth sciences such as geomorphological analysis, flood inundation mapping, hydrological analysis, large-scale mapping in an urban environment, etc., impacting the resulting output accuracy. Manual editing is done in the presented study for the automatically generated DEM from IKONOS data consequent to the satellite triangulation with a root mean square error (RMSE) of 0.46, using the rational function model (RFM) and an optimal number of ground control points (GCPs). The RFM includes the rational polynomial coefficients (RPCs) to build the relation between image space and ground space. The automatically generated DEM initially represents the digital surface model (DSM), which is used to generate a digital terrain model (DTM) in this study for improving orthoimages for an area of approximately 100 km2. DSM frequently has errors due to mass points in hanging (floating) or digging, which need correction while generating DTM. The DTM assists in the removal of the geometric effects (errors) of ground relief present in the DEM (i.e., DSM here) while generating the orthoimages and thus improves the quality of orthoimages, especially in areas such as Dehradun that have highly undulating terrain with a large number of natural drainages. The difference image of reference, i.e., edited IKONOS DEM (now representing DTM) and automatically generated IKONOS DEM, i.e., DSM, has a mean difference of 1.421 m. The difference DEM (dDEM) for the reference IKONOS DEM and generated Cartosat-1 DEM at a 10 m posting interval (referred to as Carto10 DEM) results in a mean difference of 8.74 m.

Digital elevation models (DEMs) are critical to a wide range of geoscience investigations. High-latitude and polar regions are particularly challenging for automated, stereo-photogrammetric DEM extraction due to the abundance of surfaces that are low-contrast and repetitively textured, such as snow and shadowed terrain, and have discontinuities such as in crevasse fields, glacier calving faces or iceberg edges. Sub-meter, stereo-mode satellite imagery of high geometric and radiometric quality is becoming increasingly accessible, offering the potential for dramatically increasing the spatial coverage and quality of high-latitude DEMs. Here we demonstrate and validate automated DEMs generated from the Surface Extraction with Triangulated Irregular Network-based Search-space Minimization (SETSM) algorithm designed for these challenging terrains using only the satellite rational polynomial coefficients (RPCs). Comparison between 2-m DEMs created from WorldView image pairs and low-altitude LiDAR point clouds in west Greenland give DEM biases of less than 5 m, which is the maximum systematic RPC error. Co-registration with the LiDAR data reduces the DEM RMS error to ~20 cm, which is comparable to the uncertainty of the LiDAR data. We demonstrate SETSM’s automatic RPC refinement and bias reduction by successfully extracting a high-quality DEM from Pleiades stereo pair images with large RPC errors. Finally, we provide examples of SETSM DEMs that demonstrate their utility for a range of applications of interest to polar scientists.

SAR Geocoded Ellipsoid Corrected (GEC) imagery is often taken as a post product processed from slant or ground range radar images. However, its geometric specific, being corrected to a constant ellipsoid height, makes it a coarse geocoded reference for users who are interested in accurate absolute localisation in applications, such as ‘ortho-image’ generation and digital elevation model (DEM) production. In order to improve the usefulness of SAR GEC imagery, the specific geometric distortions induced by the terrain require corresponding geometric models to perform geometric corrections to the imagery. In this article, both the rigorous physical model and the Rational Polynomial Coefficient (RPC) model for SAR GEC imagery are studied. First, the specific geometry of GEC imagery is illustrated. Then the procedure of reconstructing the rigorous physical model of GECs, which actually is mathematically combined with the normal range–Doppler model, is described in detail. The RPC estimator for replacing this rigorous physical model is then derived. Finally, based on numerous tests with the rigorous physical model available, the modelling error of the RPC is analysed. Four different kinds of high resolution SAR GEC images, i.e. TerraSAR-X, COSMO-SkyMed, RADARSAT-2 and ALOS-PALSAR, are used as experimental data. The experimental results show that using the third-order RPC model with unequal denominators as a substitute for the GEC imaging rigorous physical model, delivers accuracies for different high-resolution SAR GEC images that are all better than 0.01 pixels. The RPC model can be a robust and efficient substitute for the GEC imaging rigorous physical model to perform geometrical processing.

Abstract. The interest for the radargrammetric approach to Digital Surface Models (DSMs) generation has been growing in last years thanks to the availability of very high resolution imagery acquired by new SAR (Synthetic Aperture Radar) sensors, as COSMO-SkyMed, Radarsat-2 and TerraSAR-X, which are able to supply imagery up to 1 m ground resolution. DSMs radargrammetric generation approach consists of two basic steps, as for the standard photogrammetry applied to optical imagery: the imagery (at least a stereo pair) orientation and the image matching for the generation of the points cloud. The steps of the radargrammetric DSMs generation have been implemented into SISAR (Software per Immagini Satellitari ad Alta Risoluzione), a scientific software developed at Geodesy and Geomatics Institute of the University of Rome “La Sapienza”. Moreover, starting from the radargrammetric orientation model, a tool for the Rational Polynomial Coefficients (RPCs) for SAR images have been implemented. The possibility to generate RPCs, re-parametrizing a rigorous orientation model through a standardized set of coefficients which can be managed by a Rational Polynomial Coefficients (RPFs) model (similarly to optical high resolution imagery) sounds of particular interest since, at present, the most part of SAR imagery (except from Radarsat-2) is not supplied with RPCs, although the corresponding RPFs model is available in several commercial software. In particular the RPCs model has been used in the matching process and in the stereo restitution for the DSMs generation, with the advantage of shorter computational time. This paper discusses the application and the results of the implemented algorithm for radargrammetric DSMs generation from TerraSAR-X SpotLight imagery, acquired in Spotlight mode over Trento (Northern Italy). Urban and extra-urban (forested, cultivated) areas were considered in two different tiles, and a final overall accuracy ranging from 4.5 to 6 meters was achieved as regards the point clouds, enough well distributed independently from the land cover; moreover, it was highlighted the benefit to filter the originally derived points cloud with a global DSM as SRTM DEM, what leads to an accuracy improvement of about 20% paying a loss of matched points of about 10 %.