ZiYuan-3 Research Articles

The turning point between urban vegetation and artificial surfaces for their competitive effect on land surface temperature

Abstract As the two most common types of land cover in cities, vegetation (Veg) and artificial surfaces (AS) often exhibit competitive effects, i.e., cooling effect and warming effect, on land surface temperature (LST). Hitherto, the change of this competitive effect along the proportion gradient of AS within urban areas and their implications for urban construction still lacks adequate attention and discussion. To fill this gap, we made a quantitative analysis of the relationship between Veg (trees or grassland), AS and LST in 35 major cities of China by using Ziyuan-3 (ZY-3) high-resolution satellite observations. Results found that: (1) in each city, there exists a certain threshold (or “turning point”) along the proportion gradient of AS, exceeds which AS replaces Veg as the variable that have dominant effect on LST (i.e., the warming effect of AS is always stronger than the cooling effect of Veg); (2) for most cities, the turning points of AS for grassland and trees are 60% and 70%, respectively; (3) the turning point for cities at a higher development level is lower, indicating that even a relatively low AS coverage (∼50–60%) in these areas can lead to an evident rise in LST; 4) compared to cities in temperate and tropical climate zones, the turning point for arid/semi-arid cities is higher, implying that their urban Veg shows a better performance in mitigating urban heat stress. This study represents a systematic investigation of the competitive effect of urban Veg and AS on LST, and the understanding of turning point provides a new perspective for stakeholders to integrate urban development and temperature regulation in planning initiatives.

Examining the Roles of Spectral, Spatial, and Topographic Features in Improving Land-Cover and Forest Classifications in a Subtropical Region

Many studies have investigated the effects of spectral and spatial features of remotely sensed data and topographic characteristics on land-cover and forest classification results, but they are mainly based on individual sensor data. How these features from different kinds of remotely sensed data with various spatial resolutions influence classification results is unclear. We conducted a comprehensively comparative analysis of spectral and spatial features from ZiYuan-3 (ZY-3), Sentinel-2, and Landsat and their fused datasets with spatial resolution ranges from 2 m, 6 m, 10 m, 15 m, and to 30 m, and topographic factors in influencing land-cover classification results in a subtropical forest ecosystem using random forest approach. The results indicated that the combined spectral (fused data based on ZY-3 and Sentinel-2), spatial, and topographical data with 2-m spatial resolution provided the highest overall classification accuracy of 83.5% for 11 land-cover classes, as well as the highest accuracies for almost all individual classes. The improvement of spectral bands from 4 to 10 through fusion of ZY-3 and Sentinel-2 data increased overall accuracy by 14.2% at 2-m spatial resolution, and by 11.1% at 6-m spatial resolution. Textures from high spatial resolution imagery play more important roles than textures from medium spatial resolution images. The incorporation of textural images into spectral data in the 2-m spatial resolution imagery improved overall accuracy by 6.0–7.7% compared to 1.1–1.7% in the 10-m to 30-m spatial resolution images. Incorporation of topographic factors into spectral and textural imagery further improved overall accuracy by 1.2–5.5%. The classification accuracies for coniferous forest, eucalyptus, other broadleaf forests, and bamboo forest can be 85.3–91.1%. This research provides new insights for using proper combinations of spectral bands and textures corresponding to specifically spatial resolution images in improving land-cover and forest classifications in subtropical regions.

GEOMETRIC ACCURACY EVALUATION OF THE ZY-3 STEREO MAPPING SATELLITE FOR 8 YEARS

Abstract. The Ziyuan-3 (ZY-3) is the first civil high-resolution optical stereoscopic mapping satellite developed by China. It was launched on January 9, 2012, and as of January 9, 2020, it has been in a continuous operation for 8 years in orbit. The direct geolocation performances of ZY-3 based on the Rational Polynomial Coefficients (RPC) model are discussed, and the monoscopic and stereoscopic geometric accuracy of ZY-3 with ground control point (GCP) have also been studied using a bundle block adjustment with the affine correction in image space. Through the analysis of 47 sets of ZY-3 stereo data in Harbin, China, it can be found that from 2012 to 2019, the ZY-3 geolocation accuracy without GCP remains stable, with an average planimetric accuracy of approximately 16.2m, an average elevation accuracy of about 7.5m. The internal accuracy of the panchromatic triplet images is better than 1.0 pixel, the planimetric accuracy and elevation accuracy of stereo triplets and and stereo pairs are better than 2.0 meters with control points. In general, the results show that the ZY-3 satellite has been in good and stable condition during in-orbit operation in the past 8 years, and the various test indicators have been able to meet the design specifications.

Past and Present Practices of Topographic Base Map Database Update in Nepal

Topographic Base Maps (TBMs) are those maps that portray ground relief as the form of contour lines and show planimetric details. Various other maps like geomorphological maps, contour maps, and land use planning maps are derived from topographical maps. In this constantly changing world, the update of TBMs is indispensable. In Nepal, their update and maintenance are done by the Survey Department (SD) as a national mapping agency. This paper presents the history of topographical mapping and the reasons for the lack of updates. Currently, the SD is updating the TBM database using panchromatic and multispectral images from the Zi Yuan-3 (ZY-3) satellite with a resolution of 2.1 and 5.8 m, respectively. The updated methodology includes the orthorectification of images, the pansharpening of images, field data collection, digitization, change detection, and updating, the overlay of vector data and field verification, data quality control, and printing map production. A TBM in the Dang district of Nepal is presented as casework to show the changes in the area and issues faced during the update. Though the present digitizing procedure is time-consuming and labor-intensive, the use of high-resolution imagery has made mapping accurate and has produced high-quality maps. However, audit and automation can be introduced from the experiences of other countries for accurate and frequent updates of the TBM database in Nepal.

Sugarcane Planting Area Classification, Extraction and Accuracy Comparison Based on Chinese High-Resolution Remote Sensing Satellite Data: A Case Study of Ningming Sugarcane Demonstration Area, Chongzuo City, Guangxi

Remote sensing techniques are effective in sugarcane extraction and monitoring, but most of the existing research is based on low- and medium-resolution image. Thus, the technical methodology for high-resolution image needs to be improved. Due to the good performances of deep learning algorithms in solving classification problems for the very high resolution (VHR) images, the target mask U-Net model is introduced to research VHR satellite data from China, i. e., the GaoFen-1 (GF-1), GaoFen-2 (GF-2) and ZiYuan-3 (ZY-3). First, a sugarcane area was classified and extracted in the Ningming Sugarcane Demonstration Area in Chongzuo City, Guangxi. Further, we validated and compared the extraction accuracies for different satellite data. The results showed that the extraction accuracies of the GF-1, GF-2 and ZY-3 were 79.97% (Kappa coefficient of 0.19), 94.02% (Kappa coefficient of 0.82) and 81.94% (Kappa coefficient of 0.35), respectively. The spectral and textural information of high-resolution images can effectively guarantee improvements to the accuracy of crop extraction. By comparison of data sources and traditional supervision classification methods, the GF-2 data features the best results for sugarcane extraction. The technical methods and experimental results in this paper not only confirm the feasibility of applying China’s VHR data to monitor sugarcane planting areas, but also provides reference for the relevant future studies.

An Optimized Object-Based Random Forest Algorithm for Marsh Vegetation Mapping Using High-Spatial-Resolution GF-1 and ZY-3 Data

Discriminating marsh vegetation is critical for the rapid assessment and management of wetlands. The study area, Honghe National Nature Reserve (HNNR), a typical freshwater wetland, is located in Northeast China. This study optimized the parameters (mtry and ntrees) of an object-based random forest (RF) algorithm to improve the applicability of marsh vegetation classification. Multidimensional datasets were used as the input variables for model training, then variable selection was performed on the variables to eliminate redundancy, which improved classification efficiency and overall accuracy. Finally, the performance of a new generation of Chinese high-spatial-resolution Gaofen-1 (GF-1) and Ziyuan-3 (ZY-3) satellite images for marsh vegetation classification was evaluated using the improved object-based RF algorithm with accuracy assessment. The specific conclusions of this study are as follows: (1) Optimized object-based RF classifications consistently produced more than 70.26% overall accuracy for all scenarios of GF-1 and ZY-3 at the 95% confidence interval. The performance of ZY-3 imagery applied to marsh vegetation mapping is lower than that of GF-1 imagery due to the coarse spatial resolution. (2) Parameter optimization of the object-based RF algorithm effectively improved the stability and classification accuracy of the algorithm. After parameter adjustment, scenario 3 for GF-1 data had the highest classification accuracy of 84% (ZY-3 is 74.72%) at the 95% confidence interval. (3) The introduction of multidimensional datasets improved the overall accuracy of marsh vegetation mapping, but with many redundant variables. Using three variable selection algorithms to remove redundant variables from the multidimensional datasets effectively improved the classification efficiency and overall accuracy. The recursive feature elimination (RFE)-based variable selection algorithm had the best performance. (4) Optical spectral bands, spectral indices, mean value of green and NIR bands in textural information, DEM, TWI, compactness, max difference, and shape index are valuable variables for marsh vegetation mapping. (5) GF-1 and ZY-3 images had higher classification accuracy for forest, cropland, shrubs, and open water.

A Novel Stereo Matching Algorithm for Digital Surface Model (DSM) Generation in Water Areas

Image dense matching has become one of the widely used means for DSM generation due to its good performance in both accuracy and efficiency. However, for water areas, the most common ground object, accurate disparity estimation is always a challenge to excellent image dense matching methods, as represented by semi-global matching (SGM), due to the poor texture. For this reason, a great deal of manual editing is always inevitable before practical applications. The main reason for this is the lack of uniqueness of matching primitives, with fixed size and shape, used by those methods. In this paper, we propose a novel DSM generation method, namely semi-global and block matching (SGBM), to achieve accurate disparity and height estimation in water areas by adaptive block matching instead of pixel matching. First, the water blocks are extracted by seed point growth, and an adaptive block matching strategy considering geometrical deformations, called end-block matching (EBM), is adopted to achieve accurate disparity estimation. Then, the disparity of all other pixels beyond these water blocks is obtained by SGM. Last, the median value of height of all pixels within the same block is selected as the final height for this block after forward intersection. Experiments are conducted on ZiYuan-3 (ZY-3) stereo images, and the results show that DSM generated by our method in water areas has high accuracy and visual quality.

Glacier Mass Balance in the Nyainqentanglha Mountains between 2000 and 2017 Retrieved from ZiYuan-3 Stereo Images and the SRTM DEM

Mountain glaciers are excellent indicators of climate change and have an important role in the terrestrial water cycle and food security in many parts of the world. Glaciers are the major water source of rivers and lakes in the Nyainqentanglha Mountains (NM) region, where the glacier area has the second largest extent on the Tibetan Plateau. The potential of the high spatial resolution ZiYuan-3 (ZY-3) Three-Line-Array (TLA) stereo images to retrieve glacier mass balance has not been sufficiently explored. In this study, we optimized the procedure to extract a Digital Elevation Model (DEM) from ZY-3 TLA stereo images and estimated the geodetic mass balance of representative glaciers in the two typical areas of the NM using ZY-3 DEMs and the C-band Shuttle Radar Topography Mission (SRTM) DEM in three periods, i.e., 2000–2013, 2013–2017 and 2000–2017. The results provide detailed information towards better understanding of glacier change and specifically show that: (1) with our new stereo procedure, ZY-3 TLA data can significantly increase point cloud density and decrease invalid data on the glacier surface map to generate a high resolution (5 m) glacier mass balance map; (2) the glacier mass balance in both the Western Nyainqentanglha Mountains (WNM) and Eastern Nyainqentanglha Mountains (ENM) was negative in 2000–2017, and experienced faster mass loss in recent years (2013–2017) in the WNM. Overall, the glaciers in the western and eastern NM show different change patterns since they are influenced by different climate regimes; the glacier mass balances in WNM was –0.22 ± 0.23 m w.e. a−1 and –0.43 ± 0.06 m w.e. a−1 in 2000–2013 and 2013–2017, respectively, while in 2000–2017, it was –0.30 ± 0.19 m w.e. a−1 in the WNM and –0.56 ± 0.20 m w.e. a−1 in the ENM; (3) in the WNM, the glaciers experienced mass loss in 2000–2013 and 2013–2017 in the ablation zone, while in the accumulation zone mass increased in 2000–2013 and a large mass loss occurred in 2013–2017; as regards the ENM, the glacier mass balance was negative in 2000–2017 in both zones; (4) glacier mass balance can be affected by the fractional abundance of debris and glacier slope; (5) the glacier mass balances retrieved by ZY-3 and TanDEM-X data agreed well in the ablation zone, while a large difference occurred in the accumulation zone because of the snow/firn penetration of the X-band SAR signal.

Integrated geometric self-calibration of stereo cameras onboard the ZiYuan-3 satellite

The geometric quality of images is a principal concern for a stereo mapping satellite. On-orbit geometric calibration (GC) is an essential processing step that is always performed to ensure the geometric accuracy of satellite images. However, the traditional GC method is typically expensive and inefficient due to a large dependence on ground-based locations. Although recent GC methods based on relative constraints between images can substantially reduce plane constraints, which are used as the benchmark measurements in the traditional method, there is a strong correlation between elevation and camera orientation that necessitates additional elevation constraints. This paper proposes an integrated GC method to compensate for systematic errors in the three linear cameras (TLC) of the ZiYuan-3 (ZY-3) satellite. The relative constraints and the stereo intersection constraints from two sets of overlapped TLC imagery pairs allow for the removal of dense plane constraints, as well as elimination of the GC dependencies from the additional elevation constraints. This paper also systematically analyzes the relationship between elevation error and camera GC accuracy, proposes a unified GC model for TLC cameras based on matrix decomposition, and develops a suitable global estimation method for the solution of GC parameters. The method was validated through experiments performed on the TLC images, and satisfactory results indicated that it effectively compensates for systematic errors by onboard stereo TLC cameras.

Integrating multiview optical point clouds and multispectral images from ZiYuan-3 satellite remote sensing data to generate an urban digital surface model

The ability to accurately extract urban digital surface models (DSMs) at a reasonable cost is quite valuable. In this study, the ZiYuan-3 (ZY-3) satellite across-track and along-track modes were used to build stereo pairs and elevation models featuring different views. Image matching processing was used to generate optical point clouds for multiple stereo pairs, which were used to analyze the effects of point cloud fusion under different scenarios and optimize the ZY-3 multiview point cloud integration. Then, with the assistance of a high-precision airborne light detection and ranging point cloud, 12 indices, including the maximum, minimum, mean, and B10 to B90, were selected from the building roof optical point cloud in order to construct an urban building elevation model. The results were as follows: (1) The across-track optical point cloud formed by stereo pairs of ZY-3 nadir images is more suitable for urban DSM extraction due to the high spatial resolution and small base/height ratio of these images. (2) The B70 index performs best in the estimated urban building elevation model [ R 2 = 0.91, root mean square error (RMSE) = 5.59 m]; this model accurately identified 95.39% and 95.25% of the whole-building elevations in a narrow-scale test of 3170 buildings and a broad-scale test of 1,510,606 buildings, respectively. Moreover, the model R 2 and number of accurate building elevation recognitions gradually increased with increasing numbers of fused point clouds, while the RMSE decreased significantly. (3) A combined ZY-3 multiview point cloud fusion and multispectral data analysis technique delivered a more refined urban DSM and better characterization of building elevation under complex underlying surface scenarios than were obtained using traditional stereo pair technology. This research presents the methods and theory relating to the application of ∼2- to 6-m resolution satellite images for complex urban underlying surface scenarios.

Urban Land Use and Land Cover Classification Using Multisource Remote Sensing Images and Social Media Data

Land use and land cover (LULC) are diverse and complex in urban areas. Remotely sensed images are commonly used for land cover classification but hardly identifies urban land use and functional areas because of the semantic gap (i.e., different definitions of similar or identical buildings). Social media data, “marks” left by people using mobile phones, have great potential to overcome this semantic gap. Multisource remote sensing data are also expected to be useful in distinguishing different LULC types. This study examined the capability of combined multisource remote sensing images and social media data in urban LULC classification. Multisource remote sensing images included a Chinese ZiYuan-3 (ZY-3) high-resolution image, a Landsat 8 Operational Land Imager (OLI) multispectral image, and a Sentinel-1A synthetic aperture radar (SAR) image. Social media data consisted of the hourly spatial distribution of WeChat users, which is a ubiquitous messaging and payment platform in China. LULC was classified into 10 types, namely, vegetation, bare land, road, water, urban village, greenhouses, residential, commercial, industrial, and educational buildings. A method that integrates object-based image analysis, decision trees, and random forests was used for LULC classification. The overall accuracy and kappa value attained by the combination of multisource remote sensing images and WeChat data were 87.55% and 0.84, respectively. They further improved to 91.55% and 0.89, respectively, by integrating the textural and spatial features extracted from the ZY-3 image. The ZY-3 high-resolution image was essential for urban LULC classification because it is necessary for the accurate delineation of land parcels. The addition of Landsat 8 OLI, Sentinel-1A SAR, or WeChat data also made an irreplaceable contribution to the classification of different LULC types. The Landsat 8 OLI image helped distinguish between the urban village, residential buildings, commercial buildings, and roads, while the Sentinel-1A SAR data reduced the confusion between commercial buildings, greenhouses, and water. Rendering the spatial and temporal dynamics of population density, the WeChat data improved the classification accuracies of an urban village, greenhouses, and commercial buildings.

Estimation and analysis of along-track attitude jitter of ZiYuan-3 satellite based on relative residuals of tri-band multispectral imagery

Many high-resolution optical satellites suffer from the issue of attitude jitter that restricts the geometric quality and imaging performance. Recently, detection and estimation of spacecraft attitude jitter from imaging sensors has attracted extensive attentions because of its convenience and flexibility. Due to the existence of other impacting factors such as topographic effect, the image-based attitude jitter estimation in the along-track direction is a more challenging task and thus has been less concerned than that in the cross-track direction. This paper proposes a novel along-track jitter estimation method based on relative residuals of tri-band multispectral imagery, which effectively removes the influence of topographic effect and integration time variation. The relative residuals without the influence of topographic information are firstly acquired from a three-line parallax observation system using a precise dense matching scheme with subpixel phase correlation. The influence of integration time variation is analyzed and removed through the specific processing. Finally, the along-track attitude jitter is estimated from the relative residuals according to the harmonic addition theorem in the spatial-domain rebuilding method, and by means of Fourier transform and nonlinear optimization in the frequency-domain rebuilding method. Experiments were carried out with ZiYuan-3 (ZY-3) multispectral imagery. The comparisons with the conventional method and the attitude measurement data validate the feasibility and reliability of the proposed method, and the topographic parallax analysis after distortion correction demonstrates the performance of jitter compensation using the estimated results. In addition, a further multitemporal analysis confirms the discovery of attitude jitter with a new frequency component of around 0.2 Hz in the along-track direction and reveals the varying trends for the along-track attitude jitter of ZY-3 satellite.

Integration of ZiYuan-3 Multispectral and Stereo Data for Modeling Aboveground Biomass of Larch Plantations in North China

Data saturation in optical sensor data has long been recognized as a major factor that causes underestimation of aboveground biomass (AGB) for forest sites having high AGB, but there is a lack of suitable approaches to solve this problem. The objective of this research was to understand how incorporation of forest canopy features into high spatial resolution optical sensor data improves forest AGB estimation. Therefore, we explored the use of ZiYuan-3 (ZY-3) satellite imagery, including multispectral and stereo data, for AGB estimation of larch plantations in North China. The relative canopy height (RCH) image was calculated from the difference of digital surface model (DSM) data at leaf-on and leaf-off seasons, which were extracted from the ZY-3 stereo images. Image segmentation was conducted using eCognition on the basis of the fused ZY-3 multispectral and panchromatic data. Spectral bands, vegetation indices, textural images, and RCH-based variables based on this segment image were extracted. Linear regression was used to develop forest AGB estimation models, where the dependent variable was AGB from sample plots, and explanatory variables were from the aforementioned remote-sensing variables. The results indicated that incorporation of RCH-based variables and spectral data considerably improved AGB estimation performance when compared with the use of spectral data alone. The RCH-variable successfully reduced the data saturation problem. This research indicated that the combined use of RCH-variables and spectral data provided more accurate AGB estimation for larch plantations than the use of spectral data alone. Specifically, the root mean squared error (RMSE), relative RMSE, and mean absolute error values were 33.89 Mg/ha, 29.57%, and 30.68 Mg/ha, respectively, when using the spectral-only model, but they become 24.49 Mg/ha, 21.37%, and 20.37 Mg/ha, respectively, when using the combined model with RCH variables and spectral band. This proposed approach provides a new insight in reducing the data saturation problem.

Global Iterative Geometric Calibration of a Linear Optical Satellite Based on Sparse GCPs

Independent methods for geometric calibration (GC) have become an important research direction in the field of optical satellite technology. The main purpose of this research is to eliminate dependence on ground calibration sites using relative constraints between images. Based on a systematic analysis of these relative constraints, we found that it was difficult, if not impossible, to completely eliminate ground constraints, although the number of ground control points (GCPs) required can be greatly reduced. To achieve practical GC with high accuracy and low cost, we proposed a new method to compensate for systematic errors in linear optical satellite data acquisition using only the relative constraints between two overlapped images, namely, the corresponding elevation constraints and sparse GCPs. We first demonstrated the feasibility of GC with relative constraints and established an optimized GC model suitable for these relative constraints. We then presented a global iterative method to eliminate inaccuracies in internal calibration caused by the different distributions of GCPs within two images. The nadir (NAD) linear camera on board the Zi-Yuan 3 (ZY-3) satellite was used to evaluate the feasibility of the presented GC method; the results indicated that the present method effectively compensated for systematic errors. Thus, this article demonstrated the feasibility of GC without calibration sites.

Bundle adjustment of satellite images based on an equivalent geometric sensor model with digital elevation model

High-precision georeferencing of satellite stereo images becomes a key issue in multi-source data integration, geometric processing, and analysis for many remote sensing applications. Using existing geographic data, including readily available public data, as geometric reference for georeferencing satellite imagery is an effective and feasible way to improve the positional accuracy and reduce the costs and workforce restrictions on ground control points (GCPs) acquisition. This study proposes a novel bundle adjustment approach of push broom satellite imagery based on an equivalent geometric sensor model (EGSM). As an extension of the collinearity equations, EGSM is equivalent to the rigorous geometric sensor model, whose interior and exterior orientation parameters have clear geometric interpretations. The initial values of EGSM’s parameters can be completely recovered from the RPCs and optimized within the block adjustment without using any metadata of the satellite imagery. Furthermore, a publicly accessible digital elevation model (DEM) is used as constraints in the bundle adjustment to improve the direct georeferencing accuracies without GCPs. A set of pseudo observation equations is incorporated into the EGSM associated with appropriate weights based on the estimated variance of height differences between the tie points and the DEM surface. The performance of the proposed approach was evaluated by using 143 stereo scenes of Chinese Ziyuan-3 (ZY-3) images and the global DEM data SRTM GL3, ASTER GDEM, and AW3D30. The experimental results revealed that the proposed approach could improve not only vertical accuracies but also horizontal accuracies of satellite stereo images. The geopositional accuracy after adjustment is dependent on the quality and accuracy of the reference DEM itself. With AW3D30 as exclusive controls, the horizontal and vertical root mean square errors (RMSEs) of the experimental images are reduced from 17.3 m and 2.6 m to 2.5 m and 1.5 m, respectively. The horizontal accuracy corresponds to about 1.2 pixels in the image space, and the vertical accuracy is about 0.6 pixels with the base-height ratio of ZY-3 stereo images being 0.81. The generazation ability of the approach was further proved by the experiments of WorldView-3 stereo images.

Automatic extraction of built-up area from ZY3 multi-view satellite imagery: Analysis of 45 global cities

Accurate delineation of global built-up area (BUA) is fundamental to a better understanding of human development and the impacts on global environmental change. Existing global datasets of human settlement were mostly generated at medium and coarse spatial resolutions, including BUA and other impervious surfaces. With multiple high-resolution satellite constellations now available (e.g., ZiYuan-3 (ZY3), SPOT-5/6/7, Cartosat-1/2, and WorldView-2/3), identifying the global BUA explicitly from the complex landscapes becomes possible. In this study, a novel method was proposed for automated extraction of BUA at the global scale, by fusing a series of building features. Specifically, two planar features, the Morphological Building Index (MBI) and Harris corner detector, were employed to characterize the structure and corner attributes of buildings. Moreover, two multi-angular built-up indices (MABIs), i.e., Ratio Multi-angular Built-up Index (RMABI) and Normalized Difference Multi-angular Built-up Index (NDMABI), were proposed to represent the vertical properties of buildings based on multi-view images, which can further complement the planar features. 45 global cities were selected to validate the performance of the proposed method with images acquired by the ZY3 satellite constellation. The results show that the fusion of MBI and Harris corner can achieve satisfactory accuracy, i.e., 91.12%, 88.85%, 82.82% and 0.85, for the average overall accuracy (OA), user's accuracy (UA), producer's accuracy (PA), and F1-score, respectively, for all the test cities. After fusing the MABIs with the planar features, the average OA, UA, PA and F values of the final results were 92.00%, 86.20%, 89.14% and 0.87 for the RMABI, and 91.83%, 85.51%, 89.62% and 0.87 for the NDMABI, respectively. In particular, addition of the MABIs can further reduce the omission errors where medium/high rise buildings with low local contrast exist. We compared our results with two existing state-of-the-art global BUA products, Global Human Settlement Layer (GHSL) and Global Urban Footprint (GUF), which further corroborated the effectiveness of our method.

Mapping Forest Canopy Height in Mountainous Areas Using ZiYuan-3 Stereo Images and Landsat Data

Forest canopy height is an important parameter for studying biodiversity and the carbon cycle. A variety of techniques for mapping forest height using remote sensing data have been successfully developed in recent years. However, the demands for forest height mapping in practical applications are often not met, due to the lack of corresponding remote sensing data. In such cases, it would be useful to exploit the latest, cheaper datasets and combine them with free datasets for the mapping of forest canopy height. In this study, we proposed a method that combined ZiYuan-3 (ZY-3) stereo images, Shuttle Radar Topography Mission global 1 arc second data (SRTMGL1), and Landsat 8 Operational Land Imager (OLI) surface reflectance data. The method consisted of three procedures: First, we extracted a digital surface model (DSM) from the ZY-3, using photogrammetry methods and subtracted the SRTMGL1 to obtain a crude canopy height model (CHM). Second, we refined the crude CHM and correlated it with the topographically corrected Landsat 8 surface reflectance data, the vegetation indices, and the forest types through a Random Forest model. Third, we extrapolated the model to the entire study area covered by the Landsat data, and obtained a wall-to-wall forest canopy height product with 30 m × 30 m spatial resolution. The performance of the model was evaluated by the Random Forest’s out-of-bag estimation, which yielded a coefficient of determination (R2) of 0.53 and a root mean square error (RMSE) of 3.28 m. We validated the predicted forest canopy height using the mean forest height measured in the field survey plots. The validation result showed an R2 of 0.62 and a RMSE of 2.64 m.

Opium Poppy Detection Using Deep Learning

Opium poppies are a major source of traditional drugs, which are not only harmful to physical and mental health, but also threaten the economy and society. Monitoring poppy cultivation in key regions through remote sensing is therefore a crucial task; the location coordinates of poppy parcels represent particularly important information for their eradication by local governments. We propose a new methodology based on deep learning target detection to identify the location of poppy parcels and map their spatial distribution. We first make six training datasets with different band combinations and slide window sizes using two ZiYuan3 (ZY3) remote sensing images and separately train the single shot multibox detector (SSD) model. Then, we choose the best model and test its performance using 225 km2 verification images from Lao People’s Democratic Republic (Lao PDR), which exhibits a precision of 95% for a recall of 85%. The speed of our method is 4.5 km2/s on 1080TI Graphics Processing Unit (GPU). This study is the first attempt to monitor opium poppies with the deep learning method and achieve a high recognition rate. Our method does not require manual feature extraction and provides an alternative way to rapidly obtain the exact location coordinates of opium poppy cultivation patches.

An automatic shadow detection method for high-resolution remote sensing imagery based on polynomial fitting

ABSTRACTMost existing shadow detection models and algorithms require extensive calculations and have difficulties effectively removing features, such as water bodies, some dark objects and bluish ground objects. In this paper, we propose a high-resolution automatic shadow extraction algorithm based on the process of histogram fitting. First, the histogram of the whole image is fitted by fourth and fifth-degree polynomials according to the histogram difference of the near-infrared bands of different shadow areas in the remotely sensed image. Second, the shadow area is preliminarily extracted based on the relationships between the shadow features of the remote sensing image and the intersections of the fourth- and fifth-degree polynomials. Then, the normalized difference water index (NDWI) is applied to extract the water bodies. Finally, to obtain the shaded area, the scanning line seed filling algorithm is applied to remove the water bodies falsely detected as shadows in the preliminary shading extraction. The proposed algorithm is evaluated by using the various high-resolution images including GaoFen-1 (GF-1), GaoFen-2 (GF-2), QuickBird2, and ZiYuan-3 (ZY-3), as well as an elaborate comparison to histogram threshold segmentation algorithms such as Component 3 (C3) algorithm, multi-elements extraction algorithm multi-band detection algorithm, and spectral correlation algorithm based on spectral features. The results of experiment showed that the proposed algorithm could extract the shadows of various images, achieve satisfied results, and completely remove water bodies.

A novel change detection approach based on visual saliency and random forest from multi-temporal high-resolution remote-sensing images

ABSTRACTThis article presents a novel change detection (CD) approach for high-resolution remote-sensing images, which incorporates visual saliency and random forest (RF). First, highly homogeneous and compact image super-pixels are generated using super-pixel segmentation, and the optimal segmentation result is obtained through image superimposition and principal component analysis. Second, saliency detection is used to guide the search of interest regions in the initial difference image obtained via the improved robust change vector analysis algorithm. The salient regions within the difference image that correspond to the binarized saliency map are extracted, and the regions are subject to the fuzzy c-means (FCM) clustering to obtain the pixel-level pre-classification result, which can be used as a prerequisite for super-pixel-based analysis. Third, on the basis of the optimal segmentation and pixel-level pre-classification results, different super-pixel change possibilities are calculated. Furthermore, the changed and unchanged super-pixels that serve as the training samples are automatically selected. The spectral features and Gabor features of each super-pixel are extracted. Finally, super-pixel-based CD is implemented by applying RF based on these samples. Experimental results on Quickbird, Ziyuan 3 (ZY3), and Gaofen 2 (GF2) multi-spectral images show that the proposed method outperforms the compared methods in the accuracy of CD, and also confirm the feasibility and effectiveness of the proposed approach.