最新刊期
卷 21 ， 期 2 ， 2017
-
摘要：Snow is one of the important parts of the crysphere, because its high reflectivity and low thermal conductivity can directly affect the ground and air temperatures, albedo of the Earth’s surface, and soil moisture. In mid-latitude arid and semi-arid mountainous regions, seasonal melt are mainly supplies the spring runoff. The changes of snow distribution in mountainous regions exert an important influence on the fluctuation in river runoffs during the snowmelt season. The currently used snow products in mountainous and forested regions exhibit low accuracy when simulating mountainous hydrological processes. Thus, this research aims to develop a highly accurate snow cover mapping method for complex environments, such as mountainous and forested regions.Normalized Difference Snow Index (NDSI) based on the spectral characteristics of snow is extensively used in mapping snow cover at the global scale. However, NDSI presents a small value and discrete distribution in snow-covered forests. Thus, it cannot effectively identify snow in forested areas. Meanwhile, Normalized Difference Forest Snow Index (NDFSI) is based on the combination of the spectral features of snow and forest and defined as NDFSI=(ρnir–ρswir)/(ρnir+ρswir). In this study, a snow-cover mapping method that combines the NDSI and NDFSI was used for snow extraction in the taiga forest of Altay Mountains from the Landsat OLI image acquired in spring. First, the NDSI was used in all the pixels. The pixels with NDSI of >0.4 and ρnir of >0.11 were classified as snow, whereas those with NDSI of >0.4 but ρnir of ≤0.11 were classified as water. In this step, most snow without shielding can be recognized, and forest snow is usually not recognized because of the shielding by the forest crown. Second, NDFSI was used in pixels with NDSI of ≤0.4, and the pixels with NDFSI exceeding 0.4 were classified as forest with snow. Pixels with NDFSI of less than 0.4 were defined as forest without snow.The snow extraction result shows that snow in the no-forest region can be extracted using a reasonable NDSI threshold value. However, the snow in the forested region cannot be recognized with NDSI alone. In NDSI, the snow-covered area is underestimated in the forested-region. Thus, NDSI and NDFSI were combined, and the snow extraction result was evaluated using DEM data, temperature inversion result, and GF-1 image. Result showed that the average altitude of the forest with snow is 1611 m, and the average altitude of the forest without snow is 1278 m. The reason is that snow in lower altitudes melts earlier than snow in higher altitudes during the snowmelt season. The average temperature of forest with snow and forest without snow are 6.8 ℃ and 15.3 ℃, respectively. These temperatures are consistent with those obtained from the field. In the GF-1 image with 2 m spatial resolution can indicate the presence of snow in a forest. According to the obtained GF-1 image, the snow extraction in this study is highly accurate, and thus most snow in forest can be extracted correctly.NDFSI outperforms NDSI in the extraction of snow in forested areas. The accuracy of snow-cover mapping in complex mountainous and forested environments can be improved considerably by combining NDSI and NDFSI. Moreover, this approach can be applied easily without using other auxiliary data, such as forest maps.关键词：remote sensing;snow cover mapping;mountainous forest;NDSI;NDFSI- 417
- |
- 221
- |
- 5
发布时间：2023-06-08 -
摘要：A megacity is a city with more than 10 million inhabitants. Only two megacities existed in the 1950 s, namely, New York and Tokyo. The number of megacities eventually reached 29 in 2014. Experts predict that more than 40 megacities will exist by 2030, and at least 70% of them will come from the Maritime Silk Road. Uncontrollable urban sprawl in the Maritime Silk Road has resulted in serious environmental pollution and ecological damage, which have significantly impacted people’s daily lives and health. Building the 21st Century Maritime Silk Road is currently a hot topic. Monitoring and analyzing the spatial dynamics and patterns of megacities along the Maritime Silk Road is critical to local resources and environment protection. This study attempts to monitor and analyze the dynamics of urban expansion in the 11 megacities along the Maritime Silk Road in the period of the 1970 s to the year 2015. Five long-time series of Landsat MSS/TM/ETM+/OLI and HJ-1 CCD imagery (acquired in the 1970 s, 1990, 2000, 2010, and 2015) were adopted to extract the impervious surface of megacities along the Maritime Silk Road in this study. The images of the said megacities were geo-referenced with registration errors of less than 0.5 pixels in the data preprocessing. All images were resampled to 30 m spatial resolution under the Universal Transverse Mercator coordinates and WGS84 geodetic datum. The object-oriented Support Vector Machine (SVM) classification method was applied to all images after the data preprocessing. The classes involved in the classification maps were bare soil, impervious surface, vegetation, and water body. Approximately 200 points of each class were randomly selected as the validation points. The Overall Accuracy (OA) and Kappa values were calculated by cross-validation utilizing the Google Earth and Landsat MSS/TM/ETM+/OLI imagery. Three landscape metrics, including the largest patch index, patch density, and mean Euclidean nearest neighbor distance, were also applied to analyze and compare the spatial patterns and urbanization of the megacities along the Maritime Silk Road.Consequently, the average OA and Kappa were above 87.9% and 0.87, respectively. The proposed method could accomplish the spatio-temporal change analysis of urbanization. Moreover, the megacities along the Maritime Silk Road experienced rapid expansion in the period of the 1970 s to the year 2015. These megacities have sprawled at least four times on average with reference to the impervious surfaces of the cities in the 1970 s. In particular, Guangzhou expanded 8 times. These megacities sprawled in a concentric circle or in a “dispersion, aggregation, and re-dispersion” pattern. The urbanization of the megacities along the routes, especially in developing countries such as China and India, exhibits an increasing trend.This study offers a simple method to extract impervious surfaces on a large regional scale utilizing an object-oriented SVM classifier. The spatial expansion patterns of the megacities in developing countries along the Maritime Silk Road were analyzed utilizing applications such as spatial growth analysis and landscape metrics. The consistent monitoring of megacities in this study provides scientific data for policy makers to assess the potential impacts of urbanization in future urban planning, development activities, and population expansion.关键词：Maritime Silk Road;megacities;object-oriented support vector machine;impervious surface;landscape metrics;urbanization- 294
- |
- 172
- |
- 15
发布时间：2021-06-07 -
摘要：Matrix Doubling (MD) algorithm is an analytical solution for radiation transfer equation. MD presents relative accuracy in simulating the vegetation canopy microwave signals because the algorithm considers the multi-scattering effects. However, MD is difficult to apply in retrieving work because of its complexity. In this regard, the development of a microwave vegetation emission model, which not only takes a simple form but also considers volume scattering, is of considerable interest.In this study, we present a parameterized multi-angular model of L-band microwave emission from corn. The model is derived using the MD algorithm. First, we validated the theoretical MD algorithm with in situ data obtained from the field experiment conducted in Huailai. A good agreement with experimental data lays a solid foundation for parameterization work. To start, we simulated the corn terrain emissions in both V and H polarization at 1.4 GHz by using the theoretical MD algorithm. The incident angles were set from 5°to 59° with 2° intervals in each band. The simulations covered a broad range of corn canopy and soil dielectric. Finally, the parameterized model related the total emissivity to the optical depth of the corn and the emissivity of ground. Moreover, this model provides a series of coefficients at different incident angles. The functions between the coefficient and incident angles were provided to determine the form of the model when the angle was set.The differences between the parameterized model and the theoretical model are minimal and have RMSE in the 10–3 range of the L-band. We also validated the parameterized model using airborne PLMR observations. The simulation error in three nonconsecutive days (June 7, June 26 and August 2) is within 10000 overall. This finding shows that the parameterization model is successful in the simulation of L-band microwave emissions from corn.The parameterization model not only presents a simple form but also exhibits high accuracy by including volume scattering. Moreover, unlike other semi-empirical models, the input of the parameterization model possesses physical meaning and is not based on prior knowledge. Therefore, the parameterization model is of significant importance in retrieval work. Meanwhile, the capability to simulate corn emission from the L-band at a broad angle range improves the possibility for model applications to data from satellite. However, owing to the coarse resolution of passive microwave remote sensing, the parameterization model requires further improvement to achieve universal applicability particularly in highly heterogeneous land surfaces.关键词：passive microwave;Matrix Doubling (MD);L-band;multi-angular;parameterization;corn- 233
- |
- 173
- |
- 0
发布时间：2021-06-07 -
摘要：Airborne thermal infrared remote sensing can obtain Land Surface Temperature (LST) data at a high spatial resolution. However, the swath width of airborne stripes is commonly limited, and researchers must generally acquire multiple stripes of continuous flights to cover a larger study area. LST changes rapidly over time, and the temperature clearly varies among different stripes. Thus, the temporal normalization of LST is necessary to obtain a large scene mosaicking image of different airborne stripes. However, studies on the temporal normalization of airborne LST are still rare to date. The present study first developed a practical method based on the Diurnal Temperature Cycle (DTC) model to temporally normalize the airborne LSTs in cloud-free skies. We assumed that two factors (i.e., the typical diurnal pattern and instantaneous fluctuation from the mean atmospheric conditions) were involved in detecting LST. We then established an Improved DTC (IDTC) model that analyzes the influences of wind speed and fluctuations that cannot be measured by the DTC model. The temporal normalization of the LST requires two parts. The first aspect is the DTC variables, includingTa,tm,ts, andδT. Each land cover type has its unique DTC pattern because of its physical characteristics. Therefore,in-situ LSTs in a diurnal cycle at each ground site were employed to calculate the DTC variables through the Levenberg-Marquardt method. A look-up table (LUT) for the DTC variables of different land cover types was then constructed. The linear regression technique was subsequently utilized to evaluate the LST instantaneous fluctuation of the second part. The TASI stripes were checked pixel by pixel. The variables for each pixel were then searched from the LUT to normalize each TASI stripe.First, the DTC and IDTC models were evaluated in the temporal normalization of the TASI LST on July 10, 2012 by selecting the Heihe Watershed Allied Telemetry Experimental Research (HiWATER) experiment region as the study area. Results show that the TASI LST normalized by the IDTC model was generally more consistent with the ground measured LST than the DTC model. The accuracy of the IDTC model was approximately 0.3 K to 0.6 K higher than the DTC model. Second, to better demonstrate the performance of the IDTC model in the entire study area, the TASI LST images that had been normalized to the ASTER overpass time (12:12 local time) were up-scaled to 90 m and then compared with the ASTER LST. The data indicate that both the temporal normalization methods (DTC and IDTC models) could decrease the LST differences among different TASI stripes. The TASI LST normalized by the IDTC model was more consistent with the ASTER LST, especially in the Gobi Desert and desert steppe. Third, the temporal normalization of the ground measured LST was also implemented to test the IDTC model. It further demonstrates that the IDTC model was better than the DTC model in the temporal normalization of airborne LST.This study proposes a practical method that temporally normalizes the airborne LST based on the DTC and IDTC models to lower the temperature difference among different stripes. Results indicate that the temporal normalization based on the DTC and IDTC models can effectively obtain a large scene composed of multiple flight lines before mosaicking different stripes. The IDTC model, which involves the influences of wind speed and LST fluctuation, also performs better than the DTC model. Our present research has presented a new perspective on the temporal normalization of satellite LST.关键词：land surface temperature;temporal normalization;diurnal temperature cycle model;airborne remote sensing- 313
- |
- 163
- |
- 3
发布时间：2021-06-07 -
摘要：An assessment of the MODIS Leaf Area Index (LAI) product shows a clear focus on the consistency in value and temporal trend between remote sensing products and the “ground truth.” However, few studies have comprehensively analyzed the bias sources and quantified the contribution of different biases on the global deviation. The current study evaluates the MODIS LAI product by analyzing the MODIS LAI bias in terms of three aspects: algorithm, reflectance data, and clumping effect. We then quantify the individual influence. The bias analysis and validation of the MODIS LAI product are conducted by utilizing measured data of corn plants in a field in Huailai County, Hebei Province. Results indicate an evident underestimation of the MODIS LAI by as much as 34.14% in this area. The mean LAI values of the reference LAI and MODIS LAI are 3.53 m2/m2 and 2.33 m2/m2, respectively. The reflectance data in the bias analysis has the most significant effect on the total deviation. The bias caused by the difference between the MODIS reflectance and Landsat 8 OLI reflectance is 57.50% of the total; the clumping effect accounts for 28.33% of the total; and the bias caused by algorithm is the smallest at only 14.17% of the total. The proposed method positively influences the validation of the remote sensing product and uncertainties analysis.关键词：Leaf Area Index (LAI);MODIS LAI products;validation;bias analysis;clumping effect- 358
- |
- 155
- |
- 8
发布时间：2021-06-07 -
摘要：The Polarimetric Synthetic Aperture Radar (PolSAR) system has unique advantages in observing the Earth’s surface at different times and weather conditions. However, PolSAR data restricted to an inherent imaging mechanism are corrupted by speckle noise. The presence of speckle increases the difficulty of image understanding and decreases the accuracy of subsequent image segmentation and classification. Thus, the research on the algorithms of PolSAR image speckle reduction has important theoretical significance and practical value.A non-local filter based on the three dimensional patch matching wavelet transform (NL-3DWT) was proposed to solve the problem of preserving the structural characteristics in the despeckling of PolSAR images. First, the algorithm combined an undecimated wavelet transform with the three dimensional patch group that consists of similar patches and then applied the combination to the span (or total power) image. Local linear minimum mean square error estimation was subsequently utilized to shrink the coefficients in the wavelet domain before the inverse three dimensional wavelet transform to obtain the updated span image. Second, the edge-aligned windows selected the structural similar pixels in the updated span image. The Sigma range selected the scattering similar pixels by employing the original PolSAR data. These similar pixels were adopted to construct a pixel set to participate in the final non-local means filtering. Third, the structural similarity index based on the similar pixels set was introduced to calculate the structural preserving weight function, which was utilized in the non-local means filtering with the original PolSAR data.Two sets of PolSAR data detected by the airborne AIRSAR system were adopted to verify the effectiveness of the proposed algorithm in terms of three aspects: speckle reduction, structural features preservation, and polarimetric features preservation. Considering its relevance to this study, the NL-3DWT algorithm was compared with the classical refined PolSAR filtering method of Lee and two types of the latest PolSAR nonlocal filter methods (i.e., NL-Pretest and NL-HPP). The three methods all employed the parameters derived from their respective studies to yield convincing results.The resulting images or evaluation indexes, such as Equivalent Number of Looks (ENL), Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE), and Edge Preservation Index (EPI), show that the NL-3DWT method effectively lowered the speckle and retained edge details in the speckle reduction and structural features preservation processes. In contrast, the Refined Lee still had residual speckle. Furthermore, the NL-Pretest resulted in over-filtering that appeared as false targets, although it suppressed the speckle noise. The NL-3DWT algorithm also more effectively preserved the polarimetric scattering mechanisms than the NL-HPP in terms of the H-Alpha scatter plots in the polarimetric features preservation.The proposed method can increase the accuracy of structurally similar pixel selection, enhance the expression of image structure features, and improve the weight of image structural information in the similarity measurement between patches. Test results demonstrate that the NL-3DWT algorithm effectively lower the speckle and retains the structural characteristics and polarimetric scattering characteristics in PolSAR images. However, the algorithm complexity causes difficulties in the real-time processing of PolSAR images. Hence, studying the fast algorithm must be considered.关键词：polarimetric SAR;speckle reduction;nonlocal means;wavelet transform;structure preserving- 245
- |
- 192
- |
- 9
发布时间：2021-06-07 -
摘要：The linear detection method is usually difficult to adapt to the demand of remote sensing image processing because of it exhibits poor imaging regularity, complex background texture, and strong noises. To address these problems, we proposed a new method that possesses visual saliency in the remote sensing image and can detect principal features. First, the possibility of constructing principal lines according to the Gestalt laws and the use of extracted lines as geometric primitives were analyzed. The complex projection of the lines in the principal lines was then examined, and the definition of the principal lines was provided. Furthermore, the cumulative weight matrix of the principal line and linear statistical matrix were constructed. Meanwhile, the distribution regularity of the short line weight was studied according to the Gestalt laws, and the model of the short linear weight was constructed. Accordingly, the weight allocation pattern of the different lines in the same principal line was also discussed. Finally, detailed algorithm steps were proposed according to these analyses.The key algorithm steps were described as follows: first, the chain code marshaling algorithms were employed to extract the straight lines. Second, the accumulative weighted matrix and linear statistical matrix of the principal lines were constructed. Third, the lines were sorted on the basis of their spatial positions. Fourth, according to linear weight distribution regularity, all the lines were elected to a cumulative weight matrix according to the linear weight model and distribution rules, and the results were recorded in the linear statistical matrix. Fifth, the local maximum value of the accumulative matrix was obtained to prevent parallel overlapping among the principal lines. Sixth, constraint analysis on the continuity and purity of the accumulative weighted matrix and linear statistical matrix was conducted to prevent the appearance of false principal lines. Finally, the parameters of the principal lines were obtained according to the sorting results of the weight voting matrix and weight values. Meanwhile, the principal line was obtained through its endpoints.The results of multiple SAR and optical remote sensing satellite images with strong noises showed that the traditional line extraction method can obtain only the disordered linear information, which is not clear and useful for image processing. In this study, our proposed method obtained clear principal lines by using Gestalt law on the basis of traditional linear extraction algorithm, and the results were basically in agreement with artificial visual perception. Meanwhile, the results suggest that our algorithm is superior to the traditional cluster algorithm in terms of operation efficiency and experimental effects. The experimental results indicate the potential application of our method in various fields, such as road extraction, image matching, and object recognition. However, this method also presents several shortcomings. First, the extraction results of the principal lines rely heavily on previous results. In addition, whether the linear-weighted Gaussian model established in this study is in full compliance with the Gestalt law requires further investigation. Finally, several parameter settings are experience values acquired by a large number of experiments. Thus, we hope to achieve the adaptive processing of these parameters in our future research.关键词：principal lines;Gestalt laws;visual saliency;cumulative weight matrix;linear statistical matrix- 233
- |
- 123
- |
- 7
发布时间：2021-06-07 -
Double-population differential evolution algorithm for bilinear spectral unmixing based on FAN model
摘要：Hyperspectral unmixing plays an important role in remote sensing applications by extracting the pure constituent spectra and correspondent fractions in mixing pixels. Bilinear mixing models have recently become a hot topic in nonlinear spectral unmixing research. Nevertheless, most of the existing bilinear unmixing algorithms require prior knowledge about the endmembers, which can be regarded as supervised algorithms. This paper reports on a new unsupervised unmixing algorithm based on the FAN bilinear mixing model by employing the differential evolution (DE) algorithm (i.e., DE-FAN), which can solve the bilinear unmixing problems more efficiently and accurately compare with existing unsupervised bilinear unmixing algorithms.Considering the bilinear unmixing problem with hyperspectral imagery, the proposed algorithm makes the following improvements on the standard DE algorithm. First, the mutation factor is tuned dynamically rather than fixed during the iterative process, which can significantly enhance the neighborhood search capability. Second, considering the acceleration of the convergence rate, DE-FAN constructs a double-population (i.e., endmember population and abundance population) framework such that an alternative evolution procedure is presented. Third, the population re-initialization tactic is introduced to enhance the capability of large-scale optimization. It can decrease the number of required population individuals, thus improving the computational efficiency. It can also significantly lower the trapping risk in local optimums. Finally, a cooperative co-evolution tactic is considered. The hyperspectral image is divided into several sub-images and then processed by the DE-FAN algorithm separately. The final solution of the endmember estimation can be obtained by averaging the best solutions from the sub-images. The remaining iterations can continue by fixing the endmember variables with the final endmember solution. The final solution of abundance estimation is achieved when the stopping criteria is met.We validate the DE-FAN algorithm by adopting synthetic datasets and real airborne visible/infrared imaging spectrometer images. Different endmember numbers, signal-to-noise ratios, and maximum abundances are also considered. Several state-of-the-art algorithms (i.e., MVC-NMF, SISAL+FCLS, MU-LQM, N-FINDR+CNLS, geodesic simplex volume maximization, and PG-FAN) are compared. Experimental results conducted utilizing both synthetic and real datasets indicate that the proposed DE-FAN algorithm outperforms other algorithms by obtaining more accurate endmembers and abundance.The DE-FAN algorithm overcomes the LMM shortcomings and solves the bilinear unmixing problem. The unmixing experiments demonstrate that DE-FAN can obtain more accurate results than other classical unmixing algorithms. We will consider the influence of the penalty coefficient and attempt an adaptive method in future work. We will also extend our proposed DE-FAN to more bilinear models such as the polynomial post-nonlinear model and generalized bilinear model.关键词：hyperspectral remote sensing;spectral unmixing;bilinear mixing model;differential evolution;double-population algorithm- 329
- |
- 232
- |
- 1
发布时间：2021-06-07 -
摘要：MODIS images have important application value in the field of ground monitoring, cloud classification, and meteorological research. However, their image resolutions are still limited to a certain level because of the sensor limitations and external disturbance. This study attempts to reconstruct high-resolution MODIS images that make the edge clearer and more detailed by utilizing topic learning and the sparse representation method. The application value of existing MODIS images is then improved. A super resolution reconstruction method for MODIS images based on topic learning and sparse representation is proposed. The smoothing and texture parts of MODIS images are separated by the bilateral filtering method. The texture part is regarded as a training sample composed of several “documents”. The latent semantic features of the “document” are extracted by probabilistic Latent Semantic Analysis (pLSA) to discover the inherent “topics” of “document”. The improved K-SVD method trains several high- and low-resolution dictionary pairs that are suitable to different topics based on the aforementioned scenario, where the image blocks correspond to each topic. The probabilistic latent semantic analysis method is utilized in the reconstruction phase to adaptively select the image block topic, combine the dictionary of the corresponding topic, and reconstruct the high-resolution MODIS image through the sparse coding method. First, the MODIS image is blurred and subjected to down sampling processing in the experiment process to obtain a low-resolution image. Super resolution reconstruction is performed by utilizing different methods. The PSNR and SSIM of the original high-resolution and reconstructed images were compared utilizing different methods. Results show that the PSNR of the reconstructed image by our method is higher by approximately 1 dB and 0.5 dB than the bicubic interpolation and SCSR method, respectively. Its SSIM value is also higher than those of the other methods. The visual effects of super resolution reconstruction on the real images by different methods were compared. The experimental results show that the reconstructed images by our method have a high contrast ratio and rich texture details. The human vision is more sensitive to the image texture. This study separates the smoothing and texture parts of the MODIS image through the bilateral filter. The texture part is divided into multiple topics by probabilistic latent semantic analysis. A local adaptive super resolution method is constructed, which overcomes the problem of the adaptive selection of a reasonable dictionary according to the local characteristics of MODIS images. This process was conducted under the topic model framework combined with the improved K-SVD dictionary training methods, which train several high- and low-resolution dictionary pairs suitable to different topics. The experimental results show that the multi-dictionary reconstruction method can be utilized to represent MODIS images more sparsely and enhance the image reconstruction details. The experimental results also show that the reconstructed image is superior to the traditional method in terms of the visual effects, PSNR, and SSIM.关键词：topic learning;probabilistic latent semantic analysis;sparse representation;super resolution;MODIS image- 264
- |
- 185
- |
- 4
发布时间：2021-06-07 -
摘要：Many methods for cloud shadow detection in remote sensing have been proposed, such as utilizing the threshold and time series. Adopting the spectrum threshold can encounter many difficulties, such as the selected threshold having significant uncertainty and can be influenced by other factors. The time series method requires much time, and its process is complicated. The geometric method is particularly more accurate and practical. However, few studies have investigated the geometry method to calculate the cloud shadows on complex terrains. Therefore, this paper reports on a method that identifies cloud shadows on complex topography based on the principle of geometrical optics.When a straight line crosses a plane, it intersects with the plane at a point. However, when a straight line crosses a curved surface, it can intersect with the surface at many points. We can abstract the light through the cloud pixels as a straight line, whereas the complex terrain can be treated as a curved surface. When the light arrives on the surface by a cloud, it inevitably intersects with the surface at a certain point. This point is the shadow position casted by the cloud. We combined the previous formula to calculate the cloud shadows on the horizontal surface with DEM, which is based on the above principle of the proposed algorithm.A series of simulations and experiments determined that the proposed algorithm could properly describe the cloud shadows on the complex surface. The actual calculated and visual interpretation results had high consistency in the shadow position, and these had better fitting with the coverage by comprehensive interpretation. Theγ2 value was 0.78, whereas the RMSE was 3.49, which were much better than the results that did not consider the terrain effect. Our algorithm had a clearer advantage in terms of comprehensively identifying cloud shadows compared with the results of the threshold method based on images.An algorithm was proposed in this study based on geometry method to calculate the cloud shadows on complex terrain and overcome the existing defects of the geometry method in identifying cloud shadows. This algorithm could obtain the shadow positions of corresponding cloud pixels on a complex terrain, and the distortion of cloud shadows on complex surfaces could be properly described. However, the accuracy of the calculated results could be influenced by many factors, such as the accuracy of cloud detection and thin clouds that cannot be easily detected. However, these clouds can also cast shadows on the ground. Cloud height is an important parameter in the calculation process that largely influences the result, but the acquisition of accurate cloud heights has a certain difficulty level in remote sensing. The DEM and visible band data also have a matching problem.关键词：cloud shadow;complex terrains;DEM;geometric effelt- 309
- |
- 158
- |
- 1
发布时间：2021-06-07 -
摘要：This paper reports on a method to improve the Ratio Of an Exponentially Weighted Averages (ROEWA) edge detector, so that the improved edge detector can accurately determine the positions and directions of edges for Synthetic Aperture Radar (SAR) images. We attempt to build an optimal edge detector for SAR images to obtain better results of edge detection. The edge strength index is redefined as an inverted, signed, and normalized minimum ROEWA (IROEWA), which is utilized to quantitatively describe the phase step of edges. A new method that accurately calculates edge direction is developed based on the edge strength map from IROEWA. We can obtain the possible values of edge directions in this manner, which continuously distributes from 0 degrees to 180 degrees. Therefore, we must improve the Non-Maximum Suppression (NMS) algorithm, so that it can process sub-pixels. Finally, the improved NMS algorithm is also added into the edge detection workflow. This improved edge detection algorithm is called IROEWA & NMS. We conducted two experiments for IROEWA & NMS: one employed nature SAR images, whereas the other adopted a simulation SAR image. Experiment results show that the IROEWA & NMS outperforms the original ROEWA with watershed thresholding. The IROEWA operator is faster than the ROEWA operator under the same conditions. We applied a Receiver Operating Characteristic (ROC) curve to evaluate the IROEWA & NMS and determined that its Area Under the Curve (AUC) is 0.97570; thus, it approximates the ideal optimal detector. The detection rate at the position of the optimal point in the ROC curve of the IROEWA & NMS is as high as 0.95232, whereas the false alarm rate is as low as 0.00214. The IROEWA & NMS exhibits suitable performance on both the detection and false alarm rates. It has significant application value in several fields, such as the segmentation and edge detection for SAR images.关键词：SAR images;edge detection;ROEWA;Non-Maximum Suppression(NMS);Receiver Operating Characteristic(ROC)- 374
- |
- 63
- |
- 3
发布时间：2021-06-07 -
摘要：The traditional Bag Of Words (BOW) model disregards the scene label information of remote sensing images and ambiguity or redundancy of visual vocabularies. Hence, utilizing BOW to classify categories with similar backgrounds is unsuitable. Therefore, we propose an image scene classification algorithm based on the optimization of visual words with respect to scene label information to handle the said problem.This paper reports on an image scene classification algorithm based on the optimization of visual words with respect to scene label information. The algorithm procedure is as follows: first, images are divided into patches utilizing Spatial Pyramid Matching, and then Scale Invariant Features Transform (SIFT) features are extracted for each local image patch. These features are then clustered with K-means to form a histogram of each patch at different levels utilizing the Boiman strategy. We adopt Image Frequency as the feature selection method on visual words in each category to eliminate visual vocabulary irrelevant to a specific category and obtain a class-specific codebook. Principal Component Analysis (PCA) is then utilized to eliminate redundant visual vocabulary. Finally, we produce a mixture of class-specific histograms in each image patch at different pyramid levels and a traditional histogram with an adaptive weight. A fusion of histograms will be placed in a Support Vector Machine (SVM).We conducted experiments in this study on standard datasets of scene classification. Five experiments were conducted to demonstrate the performance of proposed algorithm. The first experiment shows that our algorithm performs better than methods that do not consider the scene label information with an increased accuracy of approximately 6 percent. The second experiment shows that the proposed method suitably performs in classifying categories with similar backgrounds and classifying error decreases in most categories. The third experiment demonstrates that the accuracy of the proposed method is higher at each pyramid level, and combined pyramids can offer even higher accuracy. The fourth experiment shows that method utilizing an adaptive weighted fusion method is more accurate than methods without. The final experiment demonstrates that the proposed algorithm performs better than other representative methods under the same conditions.This study proposes a method based on the optimization of visual words with respect to scene label information. This algorithm extracts SIFT features at different levels of pyramids combined with the Boiman strategy to generate universal histograms. DF is adopted as the feature selection method to remove visual words irrelevant to a specific category. PCA is then applied to remove redundancy and obtain class-specific codebook and histograms. Finally, a practical adaptive weighted fusion method that combines the traditional histograms of different levels with the class-specific histogram is proposed and placed in an SVM trainer and classifier. The experiment results show that the proposed algorithm suitably performs in classifying categories with similar backgrounds and displays higher stability. However, the proposed algorithm only considers one SIFT descriptor that corresponds to only one visual word. We can perform experiments on one SIFT descriptor that corresponds to several visual words and other feature selection procedures in future research.关键词：scene classification;class-specific histogram;optimization of visual words;principal component analysis;image frequency;adaptive weighted mixture- 299
- |
- 150
- |
- 4
发布时间：2021-06-07 -
摘要：Particulate matter with an aerodynamic diameter of less than 2.5 μm (PM2.5) can be transmitted for long distances and remain in the air for a long period of time, which can cause haze pollution. The accurate monitoring of PM2.5 mass concentration, especially nighttime PM2.5 mass concentration, has notable significance in ambient air quality, traffic safety, and human health. This study focuses on monitoring PM2.5 based on radiative transfer theory by utilizing nighttime radiance data collected by the Day/Night Band (DNB) of the Visible Infrared Imaging Radiometer Suite (VIIRS) aboard the Suomi National Polar-orbiting Partnership satellite. Moonlight and artificial lights are the major sources of visible light at night. Nighttime city light imagery has certain instruction ability to ambient air quality. This study adopts the Support Vector Machine (SVM) method to establish the PM2.5 mass concentration inversion model. Nighttime city light intensity was selected as the input parameters for the SVM model. Spatially and temporally paired surface PM2.5 data was selected as the output parameter. This study focuses on the moonless and cloudless nights in Beijing, China from March 2015 to May 2015. First, the contrast of the DNB images qualitatively shows that the DNB imagery is sensitive to the PM2.5 changes at night. Starting from the Beer's law, this study then establishes a link between the nighttime surface PM2.5 mass concentration and the city light intensity radiance measured by the DNB. Second, the correlation coefficient between PM2.5_rh/μ and ln(I) is calculated at each PM2.5 site following the link. Results show a negative correlation, with the largest correlation reaching –0.83 at the Dingling site. This scenario reflects a higher PM2.5 mass concentration in the surface air, and the city light radiance attenuates more in the atmospheric transmission path. Finally, the cross validation of the SVM model shows a linear correlation of 0.95 with respect to the corresponding surface observation PM2.5 mass concentration and a best-fit equation ofy=0.98x–1.82. The average absolute deviation of the SVM model and observed values is 4.89 μg·m–3, whereas the least absolute deviation is only 0.58 μg·m–3. Most of the relative deviation is less than 10%, and the minimum relative deviation is only 0.25%. Furthermore, a statistical analysis illustrates the surface PM2.5 mass concentration at the VIIRS night overpass (~2:00 AM in Beijing) time is representative of the daily-mean PM2.5 during the 3-month period. This study provides a feasible method of PM2.5 inversion utilizing DNB nighttime data. The study results indicate the accuracy of the SVM model. This model is largely significant in further filling the temporal and spatial gaps of nighttime PM2.5 monitoring, which can significantly advance the research on PM2.5 effects on the weather, environment, and human health.关键词：low light;nighttime PM2.5;VIIRS/DNB;support vector machine- 349
- |
- 171
- |
- 19
发布时间：2021-06-07 -
摘要：Harbor detection in high-resolution remote sensing images is a complex scene object detection problem, which is significant for applications such as harbor monitoring and space reconnaissance. This process is commonly requires complex feature extraction and analysis with high computational complexity. Existing works often select a smaller feature extraction region utilizing a scenario set according to experience and intuition and deal with a trade-off between computational complexity and object detection accuracy. The resolution and extraction order of features must be considered systematically to resolve this dilemma.This paper reports on a complex cue visual attention model, which is a bionic human visual attention mechanism, for harbor detection in high-resolution remote sensing images. It combines multi-scale, low-level features with high-level knowledge clues in proper resolution layers of the Gauss pyramid of the input image. This model also performs natural feature integration and comprehensive classification reasoning for harbor detection. Multi-step fast algorithms are employed to lower the computational complexity of the entire algorithm.Experiments in high-resolution optical remote sensing images from different satellites validate the proposed method. Non-harbor regions are immediately excluded from complicated feature extraction, whereas regions that most likely contain harbors are focused on for feature extraction and analysis. Thus, harbor regions are rapidly located and detected under the condition of limited computational resources. Detection effects are improved with complex feature extraction and analysis with a minimal increase in computational complexity.The complex cue visual attention model proposed in this study considers the resolution and extraction order of features systematically. It lowers the computational complexity of traditional object detection methods largely without decreasing the detection accuracy and can be applied in other complex scene object detection problems.关键词：visual attention;high resolution;remote sensing image;harbor detection;complex cues- 283
- |
- 96
- |
- 0
发布时间：2021-06-07 -
摘要：Crop biomass plays an important role in food security and global carbon cycle, and the timely and efficient monitoring of biomass is crucial for precise and reasonable agricultural management. Recently, remote sensing technique has been proven to be an effective tool for biomass estimation and it can decrease the conduct of field surveys. The European Sentinel-2A satellite was successfully launched in late June 2015.This satellite can provide high spatial resolution (10 m, 20 m, and 60 m) data freely. It uses a thirteen-band spectrum ranging from the visible region to the short-wave infrared region and thus is useful in imaging planted regions with high fragmentation. For this reason, the main objective of this paper is to explore the potential of winter wheat biomass estimation based on the new Sentinel data.In this study, 17 Vegetation Indices (VIs) based on the combinations of canopy reflectance in blue, green, red, red-edge, and near-infrared bands were first derived from the Sentinel-2A imagery in April and May 2016. The Above Ground Biomass (AGB) data collected during the same period were then used for constructing the best-fit relationships between the selected VIs and AGB. The correlation and sensitivity of the relationships between them were then analyzed. Finally, the spatial distributions of the biomass in the study area were mapped through the estimation models.All the tested VIs were nonlinearly and significantly correlated with AGB and generatedR2 ranging from 0.59 to 0.83 and RMSE ranging from 180.29 g·m–2 to 0.289.79 g·m–2. Among these VIs, the red-edge chlorophyll index exhibited superior performance on AGB estimation (R2=0.83, RMSE=180.29 g·m–2), whereas the green chlorophyll index presented the highest estimation accuracy when the red-edge bands were not available (R2=0.81, RMSE=191.15 g·m–2). The scatter-plots between the VIs and AGB showed that several VIs, such as the widely used normalized difference vegetation index, saturate at moderate-to-high biomass stages (higher than 1000g·m–2) mainly because of the strong light absorption of the red band and scattering of the near-infrared band at high LAI levels. In addition, the indices incorporated red-edge bands and thus were more closely related to the biomass compared with the original indices and were able to disrupt the saturation. Sensitivity analysis results indicated that although theR2 and RMSE values of some VIs were similar, the Vis had different sensitivities. For example, the normalized difference indices and ratio indices were more sensitive to biomass variations in the low and moderate-to-high biomass stages, respectively. On the basis of their high predictive ability, high sensitivity, and high degree of linearity, we consider the red-edge simple ratio and MERIS terrestrial chlorophyll index as a stable index for AGB estimation covering the entire growing season.Our research provides a reliable approach for winter wheat biomass estimation using the Sentinel-2A data. Given that the repeat cycle will be shortened to five days when the Sentinel-2B is launched, the Sentinel data with high spatial resolution and enhanced spectral information (including threered-edge bands) is meaningful in precision agriculture, especially in yield and production prediction.关键词：Sentinel-2;winter wheat;vegetation indices;aboveground biomass;red-edge bands- 459
- |
- 336
- |
- 50
发布时间：2021-06-07
- Postal code：100190
- Tel：010-58887052 Email：nrsb@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备20021838号-9
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0