GPU加速的高分辨率DEM图像地形特征线提取算法

随着数据采集设备的发展,数字地形分析中高分辨率数字高程模型(DEM)图像越来越普遍。目前已经存在一系列的曲线结构提取算法由于计算复杂度较高,因此在针对高分辨率DEM图像提取地形特征线时效率较低。提出一种在图形处理器(GPU)上加速Steger曲线结构提取算法的策略,利用图形处理器上计算统一设备架构(CUDA)的高度并行性来加速算法中计算密集的Hessian矩阵生成模块以及图像特征点提取模块,对于百万像素级的DEM图像该算法可以获得5倍以上的加速比。     

显著度可控的DEM地形特征线提取

目的 基于数字高程模型（DEM）的地形山脊线和山谷线提取对地形模型简化、基于样本的地形合成和地形地貌研究有重要意义,针对许多传统算法无法对所提取特征线的显著度进行方便准确的控制,以及不支持环形特征线提取的问题,提出一种新的显著度可控的DEM地形特征线提取算法。方法 首先利用全局断面扫描算法提取特征点并计算各特征点的显著度,然后根据特征点的特征方向进行特征延伸以增强特征连通性,接着采用改进的Hilditch细线化算法对特征点集合进行细线化处理,之后为相邻特征点添加特征边,构成特征图,利用环路检测与破环算法检测特征图中的环路,并破除冗余小环路,最后根据分支显著度的相似度和分支方向一致性进行特征图分解,计算分解得到特征线的显著度并筛选得到最终特征线。结果 使用真实DEM数据提取最显著的若干条特征线,与现有的基于特征显著度的地形特征线提取算法进行对比,本文算法对特征图的分解能够更准确地提取主干特征线,而基于显著度的特征线筛选控制也更加准确合理。对提出的环路检测与破环算法进行实验验证,该算法能保留大的山脊线环路,破除小的冗余环路。结论 实验结果表明,本文算法能有效实现显著度可控的山脊线和山谷线自动提取,提取结果与人眼观察结果基本一致,同时能够支持含有环形特征的地形。     

DEM保真精度定量化描述模型

等高线回放法是数字高程模型(DEM)精度评估的重要方法之一,可以用于检测地形的保真程度,但是现有定量评估指标属于相对保真精度的范畴,因此基于水系和地形的固有套合特性,提出了水系和地形套合隶属度函数,从两者的套合程度实现DEM保真精度的定量化描述.首先,根据线状要素位移准则、尺度准则、平均坡度准则、最大套合偏移量准则和目视效果准则,运用模糊数学方法建立水系和地形套合隶属度函数;然后,使用4个不同地貌类型实验区域的1∶5万地形图作为基础源数据,实验分析了DEM保真精度的评估.实验结果表明水系和地形套合隶属度函数不仅可以用于DEM保真精度的定量评估,而且可以用于DEM最佳分辨率的确定以及DEM尺度转换适宜范围的确定.     

利用星载InSAR技术提取镇江地区DEM及其精度分析

InSAR技术是目前获取高精度数字高程模型（DEM）的一种新方法。为了分析InSAR技术提取DEM的精度,首先介绍了美国航天飞机雷达SRTM DEM的精度和数据结构,然后以江苏镇江地区作为试验区,采用ERS1/2卫星影像来提取DEM,并对星载SAR提取的DEM与SRTM 3弧秒分辨率DEM的精度作了比较。 结果表明,利用星载SAR提取的DEM分辨率与SRTM 3弧秒分辨率的DEM相当,能很好地显示出地形起伏（如山脉、沟谷）的纹理特征。进一步的研究还表明,利用InSAR技术提取DEM的精度与SRTM 3 DEM之间存在5米左右的系统误差,并对产生这一系统误差的原因作了详细分析。     

星载SAR干涉技术获取DEM及其精度分析

星载合成孔径雷达干涉（InSAR）技术是一种数据覆盖范围广、廉价、高效、方便的数字高程模型（DEM）获取方法,但在地面植被覆盖广、大气水汽含量高的地区其影像相干性随时间基线的增加迅速降低;同时,SAR卫星的轨道误差也影响DEM精度。利用ERS-1/2卫星串行模式SAR数据获取镇江地区DEM,分析了轨道误差对DEM精度的影响;根据干涉相位的统计特性,从理论上给出干涉相位噪声与相干系数和视数之间的关系。实验结果表明就干涉像对的卫星轨道误差和相位噪声而言,在小区域内DEM精度优于3.5 m。     

Effect of Digital Elevation Model resolution on topographic correction of airborne SAR

Topography in high relief mountainous areas may mask the signal variation in airborne Synthetic Aperture Radar (SAR) data caused by soil moisture, surface roughness and vegetation. It also affects the quality of image calibration and registration. Good quality calibration and registration are required for the use of SAR in the estimation of soil water. To address the problem of topographic effects, the widely available standard 30m x 30m United States Geological Survey (USGS) Digital Elevation Model (DEM) has been incorporated into SAR calibration and registration programs. The topographic resolution of SAR imagery relative to the USGS DEM was examined by comparing the correlation between incident angle (theta) and SAR backscatter (sigma 0) in a high resolution DEM (mapped at 1:4800 and 1:600 from aerial photography for two small areas) to that in the USGS DEM (mapped at 1 24 000). We found that SAR resolved topographic features not resolved by the USGS DEM. Filtering and aggregation techniques were applied to reduce speckle, the apparent noise due to small topographic features resolved by SAR but not resolved by the USGS DEM, and the registration error. Increasing the filter window from 3 x 3 to 5 x 5 to 9 x 9 and the cell size from 6m x 12 m to 30m x 30m to 90m x 90m, reduced the unexplained variability in backscatter by 50%. However, there was considerable unexplained variability at all levels of filtering and aggregation. Aggregation to 90m x 90m cell size resulted in blurred or obscured surface features of hydrological interest. Filtering with a 9 x 9 window and resolution cell size of 30m x 30 m was found to be optimal in terms of the amount of variability explained and the kinds of landscape features retained. Even after applying filtering and aggregation techniques, the correlation between sigma 0 and theta for the high resolution DEM (r = 0.54) was much better than for the USGS DEM ( r = 0.36). Correction functions for the numerical estimation of terrain influence on the backscatter variation in the SAR image were derived using empirical imaging models. Topographic effects on sigma 0 were further reduced in the corrected images. However, even after correction there was considerable unexplained sigma0 variability, some of which could be attributed to major topographic features. Thus, landscape features other than theta need to be incorporated in topographic correction procedures.     

Multi-scale linkages between topographic attributes and vegetation indices in a mountainous landscape

This paper addresses a few issues that are fundamental for the understanding of vegetation-topography relations: scale dependency, seasonal variability, and importance of observing individual properties. Particularly, it uses two statistical tools - Pearson's r and Moran's I - to define relationships of several topographic attributes with the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Infrared Index (NDII), and their seasonal changes (from May to July and then September) in the Mediterranean-type landscape of the Santa Monica Mountains, California. The analyses are conducted at both the original data resolution and 20 aggregated resolutions, covering a total range of 30 m to 1500 m, so that topography-vegetation relationships can be compared at different scales. Large sample sizes have supported the significance of the following main findings for this landscape. First, elevation, slope, and southness are the most relevant primary topographic attributes among the tested attributes and their importance changes across seasons. Second, NDVI, NDII, and their seasonal variations have notably different relationships (including no relationship) with topography. Third, the observed topography-vegetation correlations (r) tend to change - typically increase - with the coarsening of spatial scale. Lastly, the spatial autocorrelation of vegetation variables and topographic attributes are often comparable, and the comparability is more apparent when topography-vegetation correlations are stronger. In all, the topography-NDVI/NDII relations defined in this paper may improve the understanding of the multi-scale and property-specific role that mountain topography plays in the formation and seasonal change of vegetation patterns.     

SRTM DEM accuracy assessment over vegetated areas in Norway

Results from the Shuttle Radar Topography Mission (SRTM) are presented. The SRTM C‐band and X‐band digital elevation models (DEMs) are evaluated with regard to elevation accuracies over agricultural fields, forest areas and man‐made features in Norway. High‐resolution digital maps and satellite images are used as background data. In general, many terrain details can be observed in the SRTM elevation datasets. The elevation accuracy (90% confidence level) of the two SRTM systems is estimated to less than 6.5 m for open agricultural fields and less than 11 m considering all land surface covers. This is better than specifications. Analysis of dense Norwegian forest stands shows that the SRTM system will produce elevation data that are as much as 15 m higher than the ground surface. The SRTM DEM products will therefore partly indicate canopy elevations in forested areas. We also show that SRTM data can be used to update older DEMs obtained from other sources, as well as estimating the volume of rock removed from large man‐made gravel pits.     

Correction of atmospheric and topographic effects for high spatial resolution satellite imagery

A method for the combined correction of atmospheric and topographic effects has been developed. It accounts for horizontally varying atmospheric conditions and also includes the height dependence of the atmospheric radiance and transmittance functions to simulate the simplified properties of a threedimensional atmosphere. A Digital Elevation Model (DEM) is used to obtain information about surface elevation, slope, and orientation. Based on the Lambertian assumption the surface reflectance in rugged terrain is calculated. The method is restricted to high spatial resolution satellite sensors like Landsat TM and SPOT HRV, since some simplifying assumptions are being made to reduce the required image processing time. The possibilities and limitations of the method are critically discussed.     

Impact of reference datasets and autocorrelation on classification accuracy

Reference data and accuracy assessments via error matrices build the foundation for measuring success of classifications. An error matrix is often based on the traditional holdout method that utilizes only one training/test dataset. If the training/test dataset does not fully represent the variability in a population, accuracy may be over – or under – estimated. Furthermore, reference data may be flawed by spatial errors or autocorrelation that may lead to overoptimistic results. For a forest study we first corrected spatially erroneous ground data and then used aerial photography to sample additional reference data around the field-sampled plots (Mannel et al. 2006 Mannel, S., Hua, D. and Price, M. 2006. A method to obtain large quantities of reference data. International Journal of Remote Sensing, 27: 623–627. [Taylor & Francis Online], [Web of Science ®] , [Google Scholar]). These reference data were used to classify forest cover and subsequently determine classification success. Cross-validation randomly separates datasets into several training/test sets and is well documented to perform a more precise accuracy measure than the traditional holdout method. However, random cross-validation of autocorrelated data may overestimate accuracy, which in our case was between 5% and 8% for a 90% confidence interval. In addition, we observed accuracies differing by up to 35% for different land cover classes depending on which training/test datasets were used. The observed discrepancies illustrate the need for paying attention to autocorrelation and utilizing more than one permanent training/test dataset, for example, through a k-fold holdout method.¹     

Impact of spatial effects on income segregation indices

Residential segregation is an inherently spatial phenomenon as it measures the separation of different types of people within a region. Whether measured with an explicitly spatial index, or a classic aspatial index, a region’s underlying spatial properties could manifest themselves in the magnitude of measured segregation. In this paper we implement a Monte Carlo simulation approach to investigate the properties of four segregation indices in regions built with specific spatial properties. This approach allows us to control the experiment in ways that empirical data do not. In general we confirm the expected results for the indices under various spatial properties, but some unexpected results emerge. Both the Dissimilarity Index and Neighborhood Sorting Index are sensitive to region size, but their spatial counterparts, the Adjusted Dissimilarity Index and Generalized Neighborhood Sorting Index, are generally immune to this problem. The paper also lends weight to concerns about the downward pressure on measured segregation when multiple neighborhoods are grouped into a single census tract. Finally, we discuss concerns about the way space is incorporated into segregation indices since the expected value of the spatial indices tested is lower than their aspatial counterparts.     

GPU加速的高精度数字地面模型建模方法

以曲面轮为基础发展的高精度曲面建模方法（HASM）可以建立具有高精度的数字高程模型,但使用该方法需要求解偏微分方程离散产生的大规模线性方程组,计算量巨大,严重制约了对大规模数据的模拟应用;而现代GPU技术的发展使GPU越来越广泛地应用于通用计算加速。为了提高HASM方法的模拟速度,把高精度曲面模拟与GPU通用技术相结合,提出了GPU加速的高精度曲面建模方法。把HASM模拟过程中的有限差分离散、离散后的大规模线性系统求解分别使用GPU进行分解,使用共轭梯度（CG）和预处理共轭梯度方法（PCG）将求解任务分解为可以并行处理的独立的多任务,使得计算任务并行化,同时并行运行大规模线程,每个线程执行一个独立的任务,充分利用了现代GPU强大的通用计算能力,并行处理以获得加速。利用并行化加速的高精度曲面建模算法使用英伟达公司的统一计算开发架构（CUDA）编程实现,GPU采用该公司的Quadro 2000。分别应用该算法进行了数值实验和实际项目区数字高程模型（DEM）模拟实验。实验结果表明,充分利用GPU的并行处理能力加速后的HASM方法,在保证达到相同曲面模拟的精度条件下,和传统的CPU方法相比,算法可以获得超过一个数量级的加速。     

Some accuracy and resolution aspects of computer vision distance measurements

The distance between an object and stereo vision sensors can be measured using image processing and known system parameters. A detailed distance measurement synthesis procedure to meet system specifications is presented and illustrated with an example. An error analysis shows that error is proportional to distance. System parameters such as separation between sensor elements, sensor focal length, and sensor array dimensions are related in the design and error equations presented. The main desired design goal is to establish the smallest image sensor array size which will meet system operating specifications. Minimum and maximum distance, object height, optic parameters, scene shift, and sensor array parameters are related.     

Multivariate forest structure modelling and mapping using high resolution airborne imagery and topographic information

Remote sensing has been widely used for modelling and mapping individual forest structural attributes, such as LAI and stem density, however the development and evaluation of methods for simultaneously modelling and mapping multivariate aspects of forest structure are comparatively limited. Multivariate representation of forest structure can be used as a means to infer other environmental attributes such as biodiversity and habitat, which have often been shown to be enhanced in more structurally diverse or complex forests. Image-based modelling of multivariate forest structure is useful in developing an understanding of the associations between different aspects of vertical and horizontal structure and image characteristics. Models can also be applied spatially to all image pixels to produce maps of multivariate forest structure as an alternative to sample-based field assessment. This research used high spatial resolution multispectral airborne imagery to provide spectral, spatial, and object-based information in the development of a model of multivariate forest structure as represented by twenty-four field variables measured in plots within a temperate hardwood forest in southern Quebec, Canada. Redundancy Analysis (RDA) was used to develop a model that explained a statistically significant proportion of the variance of these structural attributes. The resulting model included image variables representing mostly within-crown and within-shadow brightness variance (texture) as well as elevation, taken from a DEM of the study area. It was applied spatially across the entire study area to produce a continuous map of predicted multivariate forest structure. Bootstrapping validation of the model provided an RMSE of 19.9%, while independent field validation of mapped areas of complex and simple structure showed accuracies of 89% and 69%, respectively. Multiscale testing using resampled imagery suggested that the methods could possibly be used with current pan-sharpened, or future sub-metre resolution, multispectral satellite imagery, which would provide much greater spatial coverage and reduced image processing compared to airborne imagery.     

Impact of product configuration systems on product profitability and costing accuracy

This article aims at analyzing the impact of implementing a product configuration system (PCS) on the increased accuracy of the cost calculations and the increased profitability of the products. Companies that have implemented PCSs have achieved substantial benefits in terms of being more in control of their product assortment, making the right decisions in the sales phase and increasing sales of optimal products. These benefits should have an impact on the company’s ability to make more accurate cost estimations in the sales phase, which can positively affect the products’ profitability. However, previous studies have not addressed this relationship to a great extent. For that reason, a configure-to-order (CTO) manufacturing company was analyzed. A longitudinal field study was performed in which the accuracy of the cost calculations and the products’ profitability were analyzed before and after a PCS was implemented. The comparison in the case study revealed that increased accuracy of the cost calculations in the sales phase and consequently increased profitability can be achieved by implementing a PCS.     

面向对象滑坡信息提取中DEM空间分辨率影响分析

数字高程模型(DEM)是识别滑坡的重要特征,而在实际应用中通常难以获取研究区的高分辨率DEM数据。为分析DEM分辨率对滑坡提取的影响并确定满足应用的分辨率要求,将原始10 m分辨率DEM重采样为空间分辨率不同的15、20、30、50、75与100 m,采用DEM及其衍生数据辅助高分遥感影像的面向对象滑坡信息提取方法进行实验。实验结果表明:当DEM分辨率小于30 m时,对面积大于5 000 m²的滑坡能得到较好的识别和分类;大于30 m时虽难以区分出滑坡类型,但通过调整部分规则参数仍能实现对滑坡的监测。该研究对于准确提取滑坡信息所需DEM产品的选择具有一定的指导意义和参考价值。     

Digital elevation model (DEM) generation and accuracy assessment from ASTER stereo data

Digital elevation models (DEM) are the indispensable quantitative environmental variable in most of the research studies in remote sensing. The improvement of sensor and satellite imaging technologies enabled the researchers to generate DEM using remotely sensed data. These data can be started to use as not only the two-dimensional (2-D) but also three-dimensional (3-D) information sources with usage of the DEM. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) is one of the sensor systems capable of DEM generation and during the study, ASTER level 1A (L1A) data were used. Due to presence of many geological features and different landcover types, the test site is selected as the watershed of Asarsuyu River, located in north-western Anatolia in between Duzce and Bolu plains. The aim of this study is to check the best effort of 15 m spatial resolution DEM generation from ASTER L1A data by collecting different numbers of ground control points (GCP) (30, 45, and 60) and tie points (TP). During the study, three different techniques—spatial correlation, image differencing and profiling—were used for both planimetric and vertical accuracy assessment. The obtained results from both of the techniques show that the accuracy of the DEM increases by increasing the number of GCP. However, there is an only slight difference between the result of 45 GCPs and 60 GCPs.     

Spatial and temporal accuracy of coarse resolution products of NOAA-AVHRR NDVI data

Coarse resolution products of maximum value composite NDVI data, derived from the AVHRR on board the NOAA-7, -9 and -11 series of polar-orbiting satellites, are becoming increasingly available. One such product is the ARTEMIS African decadal NDVT data set, available on CD-ROM for the years 1981 to 1991. These data have inherent spatial and temporal errors which arise as a result of the sampling procedures involved in their generation. This paper describes the way in which the ARTEMIS NDVI data are produced. It then describes the spatial and temporal accuracy of the data, estimated for a study area in East Africa, in terms of: 1. their spatial resolution, 2. pixel locational errors, 3. sampling biases and systematic errors introduced during the production of the data, and 4. the presence of signal noise in multi-temporal profiles of the data. The paper concludes that these errors may prove a considerable limitation to the usefulness of the data for distinguishing specific habitats on the ground over small study areas, and that in the further development of applications for these coarse resolution data sets, these errors must be taken into consideration.     

Comparing the accuracy of estimated terrain elevations across spatial resolution

Terrain is modelled in Geographic Information Science on a grid, assuming that elevation values are constant within any single pixel of a Digital Elevation Model (DEM). Pixels are considered flat and rigid, for computational simplicity (a ‘rigid pixel’ paradigm). This paradigm does not account for the slope and curvature of terrain within each pixel, generating imprecise measurements, particularly as pixel size increases or in uneven terrain. This paper relaxes the rigid pixel assumption, allowing for possible sub-pixel variations in slope and curvature (a ‘surface-adjusted’ paradigm). This paper compares different interpolation methods to investigate sub-pixel variations for estimating elevation of arbitrary points given a regular grid. Tests interpolating elevation values for 20,000 georeferenced off-centroid random points from a regular grid DEM are presented, using a variety of exact and inexact local deterministic interpolation methods within contiguity configurations incorporating first and second order neighbours. The paper examines the accuracy of surface-adjusted estimations across a progression of spatial resolutions (10 m, 30 m, 100 m, and 1,000 m DEMs) and a suite of terrain types varying in latitude, altitude, slope, and roughness, validating off-centre estimates against direct elevation measurements on 3 m resolution lidar DEM. Results illustrate that the Bi-quadratic and Bi-cubic interpolation methods outperform Weighted Average, Linear, and Bi-linear methods at coarse resolutions and in rough or non-uniform terrain. In smooth or flat terrain and at finest resolutions, the interpolation method impacts estimation accuracy less or not at all. The type of contiguity configuration appears to play a role in estimation errors as well, with tighter neighbourhoods exhibiting higher accuracy. The analysis also examined regularized mathematical surfaces, adding autocorrelated randomly distributed noise to simulate terrain. The results of experiments based on regularized smooth mathematical surfaces do not translate directly to terrain modelling. The analysis also considers the balance between the increased computation times needed to measure surface-adjusted elevation against improvements in accuracy.