Accuracy analysis, dem generation and validation using russian tk-350 stereo-images

TK-350 stereo-scenes of the Zonguldak testfield in the north-west of Turkey have been analysed. The imagery had a base-to-height ratio of 0·52 and covered an area of 200 km × 300 km, with each pixel representing 10 m on the ground. Control points digitised from 1:25 000 scale topographic maps were used in the test. A bundle orientation was executed using the University of Hanover program BLUH and PCI Geomatica OrthoEngine AE software packages. Tests revealed that TK-350 stereo-images can yield 3D geopositioning to an accuracy of about 10 m in planimetry and 17 m in height. A 40 m resolution digital elevation model (DEM) was generated by the PCI system and compared against a reference DEM, which was derived from digitised contour lines provided by 1:25 000 scale topographic maps. This comparison showed that accuracy depends mainly on the surface structure and the slope of the local terrain. Root mean square errors in height were found to be about 27 and 39 m outside and inside forested areas, respectively. The matched DEM demonstrated a systematic shift against the reference DEM visible as an asymmetric shift in the frequency distribution. This is perhaps caused by the presence of vegetation and buildings.     

Exploring The Generation Of Digital Elevation Models From Same-Side Ers Sar Images: Topographic And Temporal Effects

This paper presents an attempt to explore the generation of digital elevation models (DEMs) by radargrammetry from same-side ERS-1/2 synthetic aperture radar (SAR) images. Two ERS-1 and two ERS-2 C-band (wavelength 5·6 cm) SAR images along parallel descending orbits, acquired in 1996 and 1997, are used to form four stereopairs. The test area is western Hong Kong which is dominated by mountains/hills that are largely covered by evergreen-broadleaf trees. The stereopairs covering the same area but with different time intervals (varying from 16 days to 1·5 years) have an intersection angle of 4·5° and a relatively small overlap of 30% (30 km). In addition to the effect of topography, the effect of temporal correlation on the accuracy of the resulting DEMs is also analysed for the first time. From the experimental results obtained in this study, the following conclusions may be drawn: (a) with same-side ERS-1/2 SAR images, an accuracy of 30 to 35 m is achievable; (b) the accuracy of a resulting DEM degrades almost linearly with an increase in slope when the terrain slope is below 30°, and in the areas with slope over 30°, the accuracy of the DEM is significantly degraded and even becomes unacceptable; (c) the worst accuracy is obtained in the foreslope directions and the best accuracy is obtained in the backslope directions; (d) the time interval between two SAR images has no significant effect on the accuracy of the DEM in general; (e) the time interval between two SAR images has no significant effect on the reliability of image matching.     

Spot Revisited: Accuracy Assessment, Dem Generation and Validation from Stereo Spot 5 Hrg Images

SPOT 5 HRG Level 1A and 1B stereo scenes covering Zonguldak testfield in north-west Turkey have been analysed. They comprise the left and right image components with base to height ratio of 0·54. The pixel size on the ground is 5 m. The bundle orientation was executed by the PCI Geomatica V9.1.4 software package and resulted in 3D geopositioning to sub-pixel accuracies in each axis provided that at least six control points were used in the computation. Root mean square error (rmse) values and vectors of residual errors for Levels 1A and 1B are similar, even for different control and check point configurations. Based on the scene orientation, Level 1A and 1B digital elevation models (DEMs) of the testfield have been determined by automatic matching and validated by the reference DEM digitised from the 1:25 000 scale topographic maps, interferometric DEMs from Shuttle Radar Topography Mission (SRTM) X- and C-band SAR data and the GPS profiles measured along the main roads in the testfield. Although the accuracies of reference data-sets are too similar to the generated SPOT DEMs, these are the only high quality reference materials available in this area. Sub-pixel height accuracy was indicated by the comparison with profile points. However, they are in favourable locations where matching is always successful, so such a result may give a biased measure of the accuracy of the corresponding DEMs.     

Digital close range photogrammetry for measurement of soil erosion

Many of the processes involved in soil erosion have dimensions on the millimetre scale. Modelling and quantification of such processes require information on soil surface topography with adequate resolution. The purpose of this study was to generate digital elevation models (DEMs) from soil surfaces with high spatial and temporal resolution. Digital photogrammetry was applied for measuring erosion rates on complex-shaped soil surfaces under laboratory rainfall conditions. A total of 60 DEMs were generated, covering a planimetric area of 16 m². The DEMs had a grid resolution of 3 mm. A vertical precision of approximately 1 mm was desired for DEM analysis. A consumer-grade digital camera was used for image acquisition. The camera was calibrated using BLUH software. Homologous points in overlapping images were identified with least squares matching software. Irregularly spaced object coordinates were interpolated to a regular grid in a geographic information system. The resulting DEMs represented the soil surface well. A precision of 1·26 mm in the vertical was attained. The precision of DEM production was limited to camera calibration. Improvements of the setup presented could include the use of better control points and more advanced image matching strategies for identification of homologous points. The DEMs allowed for detailed analysis of soil surface evolution.     

Stereo-Photogrammetric Mapping of Tooth Replicas Incorporating Texture

Commercial digital photogrammetric software has been applied to convergent stereoscopic photography of human tooth replicas prepared to exhibit optical texture resulting in successful generation of 3D coordinate data. Tooth replicas were imaged using a semi-metric 35 mm camera and f = 100 mm macro lens on extension bellows. Model precision was within acceptable limits of 12  μ m or better for manual target matching and 21  μ m or better for automatic image matching. Further improvement in optical texture is required to achieve automatic image matching precision comparable to that of manual target matching. Small errors in interior orientation parameters attributed to instability in the bellows as well as small errors in the relative orientation resulted in some systematic errors. The use of a fixed camera lens system is expected to reduce these errors. When combined with commercially available, moderately priced, digital SLR cameras this brings 3D model generation closer to everyday clinical dental practice.     

Mapping tooth surface loss with a fixed-base stereo camera

Change detection of tooth surface loss in the cervical region of a tooth replica has been performed using a fixed-base stereoscopic camera, consisting of two SLR digital cameras with lenses of fixed focal length. Commercial digital photogrammetric and calibration software was used to perform a convergent multi-station calibration with an artefact comprising a transilluminated planar array and a tiered object exhibiting optical texture. Digital surface models of a tiered test object and a tooth surface containing a non-carious cervical lesion were generated from stereo-imagery. The accuracy of automatic mass point measurement of a planar surface was 4 ± 13 μm in Z. The sensitivity of change detection on mapped tooth replicas was 0·03 mm with change ranging from 0·03 to 0·07 mm per annum. Different rates of change were clearly evident in different areas at different times. The results of this investigation suggest that annual change detection studies will provide a clearer picture of the pattern of tooth surface loss and, in combination with other analytical techniques, a more detailed explanation of the natural history of these lesions.     

一种综合利用像方和物方信息的多影像匹配方法(英文)

A novel method for multi-image matching by synthesizing image and object-space information is proposed. Firstly, four levels of image pyramids are generated according to the rule that the next pyramid level is generated from the previous level using the average gray values of the 3 by 3 pixels, and the first level of pyramid image is generated from the original image. The initial horizontal parallaxes between the reference image and each searching image are calculated at the highest level of the image pyramid. Secondly, corresponding image points are searched in each stereo image pair from the third level of image pyramid, and the matching results in all stereo pairs are integrated in the object space, by which the mismatched image points can be eliminated and more accurate spatial information can be obtained for the subsequent pyramid image matching. The matching method based on correlation coefficient with geometric constraints and global relaxation matching is introduced in the process of image matching. Finally, the feasibility of the method proposed in this paper is verified by the experiments using a set of digital frame aerial images with big overlap. Compared with the traditional image matching method with two images, the accuracy of the digital surface model (DSM) generated using the proposed method shows that the multi-image matching method can eliminate the mismatched points effectively and can improve the matching success rate significantly. Supported by the National Natural Science Foundation of China (Nos. 40771176, 40721001).     

遗传算法与最小二乘法相结合的遥感图像子像素匹配

图像匹配在计算机视觉和数字摄影测量中都是一项十分重要的任务。为了加速匹配并为高精度子像素匹配提供稳健而相对精度的初值，本文提出了一种基于遗传算法（GA）和最小二乘法匹配（LSM）相结合的遥感图像子像素匹配算法。实验表明：基于遗传算法的匹配算法比序贯相似度度量（SSDA）和经典的平均绝对差（MAD）方法要快得多。以我们的实验数据为例，MAD与GA在相关计算总量上的比至少为36：1。本文从理论上证明了这一点。     

数字地形模型自动获取的一种新方案——辐射法

利用数字影像自动提取数字地形模型(DTM)的关键是影像匹配。匹配算法有基于特征的相关匹配和基于影像区域匹配两类,前者利用某种抽象的“特征”(一般为线段或点),而后者直接比较像素。但要同时满足DTM的精度,密度、可靠性和速度等要求,单纯引用某种特定的相关方法是不能有效地解决生产实际问题的。本文在分析现有几种典型方法的基础上,提出了基于特征点的辐射法,从而较为合理地兼顾了精度、密度、可靠性以及速度等指标。在AST 386微机上,无任何加速处理器,运行速度达到每秒1～3点,精度为子像素精度。     

基于特征点的影像视差分析

本文介绍了一种基于特征点的数字立体影像匹配方法。该法采用Frstner算子提取特征点,然后应用一种松驰迭代算法,实现了同名特征点之间的立体匹配。为实现立体像对的自动相对定向和数字地形模型自动生成提供了一种切实可行的方法。     

High-accuracy photogrammetric technique for human spine measurement

Close range photogrammetry has been recognised as an essential tool for the capture of high-accuracy spatial data for medical applications, in particular work involving dynamic human body parts such as limbs. Offline and online photogrammetric systems are readily available for a number of common applications. However, off-the-shelf systems are not always appropriate because of project site conditions. To achieve high measurement accuracy in a field environment, a modified field camera calibration technique was introduced. The technique is particularly important where each camera is limited to one captured image during calibration, as the camera and the calibration testfield are in fixed positions. In this paper a custom-built imaging system designed for the study of the human spine in an outdoor environment is introduced. The discussion addresses: (1) imaging system design; (2) modified field calibration techniques; and (3) a case study on human spines. Two field camera calibration techniques were evaluated, both of which improved the accuracy of the prototype system, the use of a detachable target board offering the best results. This modified camera calibration procedure has improved the 3D measurement accuracy from 1·25 ± 0·3 mm to 0·43 ± 0·1 mm. The improvement is at a level achievable in the laboratory. The technique is considered to provide accurate and reliable anthropometric landmark measurement at low cost. This was evaluated in a clinical setting where diurnal changes in spine length and contour were measured in a cohort of 30 university students. The capability of the technique to measure sagittal and frontal angular changes provides a novel way of examining changes in spine shape.     

数字影像匹配质量的一种自动诊断方法

利用欲诊断的影像匹配结果（ＤＥＭ），由左、右片影像自动生成数字正射像片对和立体正射像片对，然后用影像匹配方法自动量测其高程和高程改正值，从而自动诊断并自行改正原来影像的匹配结果，同时提供ＤＥＭ、数字正射影像和立体正射像片对３种产品。     

Towards 3D map generation from digital aerial images

This paper describes the fusion of information extracted from multispectral digital aerial images for highly automatic 3D map generation. The proposed approach integrates spectral classification and 3D reconstruction techniques. The multispectral digital aerial images consist of a high resolution panchromatic channel as well as lower resolution RGB and near infrared (NIR) channels and form the basis for information extraction.Our land use classification is a 2-step approach that uses RGB and NIR images for an initial classification and the panchromatic images as well as a digital surface model (DSM) for a refined classification. The DSM is generated from the high resolution panchromatic images of a specific photo mission. Based on the aerial triangulation using area and feature-based points of interest the algorithms are able to generate a dense DSM by a dense image matching procedure. Afterwards a true ortho image for classification, panchromatic or color input images can be computed.In a last step specific layers for buildings and vegetation are generated and the classification is updated.     

A combined image matching method for Chinese optical satellite imagery

Image matching is one of the key technologies for digital Earth. This paper presents a combined image matching method for Chinese satellite images. This method includes the following four steps: (1) a modified Wallis-type filter is proposed to determine parameters adaptively while avoiding over-enhancement; (2) a mismatch detection procedure based on a global-local strategy is introduced to remove outliers generated by the Scale-invariant feature transform algorithm, and geometric orientation with bundle block adjustment is employed to compensate for the systematic errors of the position and attitude observations; (3) we design a novel similarity measure (distance, angle and the Normalized Cross-Correlation similarities, DANCC) which considers geometric similarity and textural similarity; and (4) we introduce a hierarchical matching strategy to refine the matching result level by level. Four typical image pairs acquired from Mapping Satellite-1, ZY-1 02C, ZY-3 and GeoEye-1, respectively, are used for experimental analysis. A comparison with the two current main matching algorithms for satellite imagery confirms that the proposed method is capable of producing reliable and accurate matching results on different terrains from not only Chinese satellite images, but also foreign satellite images.     

基于深度学习的立体影像密集匹配方法综述

三维重建可用于数字高程模型制作、机器人导航、增强现实和自动驾驶等.视差图是三维重建中一种重要的表达方式,而立体密集匹配是使用最广泛的获取视差图的技术.近年来,随着硬件、数据集、算法的发展,基于深度学习的立体匹配方法受到了广泛关注并取得了巨大成功.然而,这些方法通常在近景立体像对中进行测试,很少被用于遥感影像中.回顾了双...     

无线电高度与数字地图辅助导航算法研究

无线电高度表和数字地图的组合构成地形轮廓匹配 (TERCOM )辅助导航系统 ,它是一种飞行导航定位的重要手段 ,由于地形辅助导航不依赖卫星导航系统、不受外界干扰、具有自主导航的功能 ,故近年来获得了迅速发展。本文对传统的地形匹配算法进行了如下改进 :利用实测高程序列和预测高程序列与基准图分别进行匹配 ,对飞行器进行实时定位并对下一时刻的航迹进行预测 ;利用递推和分层匹配策略提高配准效率 ;利用多判据组合匹配提高定位的可靠性。根据本文所提出的数学模型 ,并用外场飞行数据进行了匹配测试 ,结果证明该方法提高了地形轮廓匹配的可靠性和精度 ,取得了很好的效果。     

移动高程平面约束的多视影像特征点匹配

针对影像匹配中由于重复纹理和遮挡导致的不确定性问题,提出移动高程平面约束的多视影像特征点匹配方法。该方法将多视影像上提取的特征点投影到不同高程的物方格网平面,利用平面上格网单元来约束候选同名特征点,对多视影像进行选择性匹配,避免部分影像遮挡对匹配的影响。实验选取UCX(UltraCamX)数字航空影像进行匹配,取得可靠的匹配结果,从而验证了本文方法的有效性。     

Effects of jpeg 2000 compression on automated dsm extraction: evidence from aerial photographs

This paper evaluates the effects of JPEG 2000 compression on automated digital surface model (DSM) extraction, using the area-based matching technique. Two stereopairs of aerial photographs, which have different photo scale, pixel size, and topographic type, are used as the images for the experiments. In the experiments, the DSM generated from the uncompressed image is set as the reference, and the elevation accuracy is computed for a range of compression ratios (2:1 to 100:1). The results show that the employed indices of image quality based on the JPEG 2000 compression are clearly superior to the commonly used JPEG compression technique—especially for high compression ratios. However, the elevation accuracy varies with the compression ratios. Also, JPEG 2000 compression does not have a significant influence on the percentage of the successful matching rate. In addition, a near linear fall-off is observed in extracted DSM accuracy with increasing compression ratios.