Pub Date : 2022-10-20DOI: 10.1080/07038992.2022.2128731
Qin Nie, K. Shi, Xuewen Wu
Abstract An understanding of the multifractal and long-range dependence of impervious surfaces (IS) spatiotemporal patterns is helpful for regional environmental assessment and urban planning. Linear spectral-mixture analysis has been applied to compute IS in the coastal city of Xiamen, China, based on Landsat TM/OLI/TIRS images, and then the long-term trends and multifractal characteristics of IS patterns have been investigated using two-dimensional multifractal detrended fluctuation analysis. The IS spatial distribution displayed similar positive long-range correlations in study areas during the 1994–2015 period. IS have a long-memory characteristic within a certain spatial range, with an increase in the value of a pixel likely to cause an increment in the value of its neighbors. The multifractality of the IS distribution increased in Xiamen City during 1994–2015, but was lower than those in Xiamen Island. The multifractal spectra in Xiamen City vary in shape between years, capturing its evolution from right truncation with a long left tail to left truncations and long right tails and then a symmetrical shape. The fractal structure for Xiamen Island exhibits similar patterns of long right tails and left truncations. Economic and political considerations coupled with natural geographic conditions dominate the long-range trends in the IS spatial patterns.
{"title":"Multifractal and Long-Term Memory of Impervious Surface Spatial Patterns in a Coastal City in China","authors":"Qin Nie, K. Shi, Xuewen Wu","doi":"10.1080/07038992.2022.2128731","DOIUrl":"https://doi.org/10.1080/07038992.2022.2128731","url":null,"abstract":"Abstract An understanding of the multifractal and long-range dependence of impervious surfaces (IS) spatiotemporal patterns is helpful for regional environmental assessment and urban planning. Linear spectral-mixture analysis has been applied to compute IS in the coastal city of Xiamen, China, based on Landsat TM/OLI/TIRS images, and then the long-term trends and multifractal characteristics of IS patterns have been investigated using two-dimensional multifractal detrended fluctuation analysis. The IS spatial distribution displayed similar positive long-range correlations in study areas during the 1994–2015 period. IS have a long-memory characteristic within a certain spatial range, with an increase in the value of a pixel likely to cause an increment in the value of its neighbors. The multifractality of the IS distribution increased in Xiamen City during 1994–2015, but was lower than those in Xiamen Island. The multifractal spectra in Xiamen City vary in shape between years, capturing its evolution from right truncation with a long left tail to left truncations and long right tails and then a symmetrical shape. The fractal structure for Xiamen Island exhibits similar patterns of long right tails and left truncations. Economic and political considerations coupled with natural geographic conditions dominate the long-range trends in the IS spatial patterns.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"814 - 825"},"PeriodicalIF":2.6,"publicationDate":"2022-10-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49052541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-13DOI: 10.1080/07038992.2022.2130742
Martin Queinnec, N. Coops, J. White, V. Griess, N. Schwartz, G. McCartney
{"title":"Mapping Dominant Boreal Tree Species Groups by Combining Area-Based and Individual Tree Crown LiDAR Metrics with Sentinel-2 Data","authors":"Martin Queinnec, N. Coops, J. White, V. Griess, N. Schwartz, G. McCartney","doi":"10.1080/07038992.2022.2130742","DOIUrl":"https://doi.org/10.1080/07038992.2022.2130742","url":null,"abstract":"","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"1 1","pages":""},"PeriodicalIF":2.6,"publicationDate":"2022-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43009394","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-26DOI: 10.1080/07038992.2022.2121271
Laura Dingle Robertson, H. Mcnairn, X. Jiao, Connor McNairn, S. Ihuoma
Abstract The RADARSAT Constellation Mission (RCM) can acquire imagery in Compact Polarimetric (CP) mode. With this new mode, and the increased revisit with three satellites, RCM can contribute to operational crop monitoring at national scales. The four Stokes (S0, S1, S2 and S3) and three m-chi decomposition (surface, double bounce, volume) parameters were used to identify crops (pasture/forage, barley, wheat, canola, flaxseed, peas, lentils) with a Random Forest classifier. The Stokes and m-chi parameters delivered maps of similar accuracies (95% overall accuracy) and were only slightly less accurate than a classification using optical satellite imagery (97%). To understand why Stokes parameters worked well in classifying crops, scattering responses for wheat, canola, lentils and peas were plotted on the Poincaré sphere. These responses were interpreted in the context of the degree of polarization and were related to crop phenology. These plots revealed that early and late in the season the polarized component of the scattered wave remained circular. However, in the active season when crop structure was changing, scattered waves became more elliptically polarized. Although the amount of polarized scattering was lower mid-season, the change in ellipticity was helpful in separating crop types.
{"title":"Monitoring Crops Using Compact Polarimetry and the RADARSAT Constellation Mission","authors":"Laura Dingle Robertson, H. Mcnairn, X. Jiao, Connor McNairn, S. Ihuoma","doi":"10.1080/07038992.2022.2121271","DOIUrl":"https://doi.org/10.1080/07038992.2022.2121271","url":null,"abstract":"Abstract The RADARSAT Constellation Mission (RCM) can acquire imagery in Compact Polarimetric (CP) mode. With this new mode, and the increased revisit with three satellites, RCM can contribute to operational crop monitoring at national scales. The four Stokes (S0, S1, S2 and S3) and three m-chi decomposition (surface, double bounce, volume) parameters were used to identify crops (pasture/forage, barley, wheat, canola, flaxseed, peas, lentils) with a Random Forest classifier. The Stokes and m-chi parameters delivered maps of similar accuracies (95% overall accuracy) and were only slightly less accurate than a classification using optical satellite imagery (97%). To understand why Stokes parameters worked well in classifying crops, scattering responses for wheat, canola, lentils and peas were plotted on the Poincaré sphere. These responses were interpreted in the context of the degree of polarization and were related to crop phenology. These plots revealed that early and late in the season the polarized component of the scattered wave remained circular. However, in the active season when crop structure was changing, scattered waves became more elliptically polarized. Although the amount of polarized scattering was lower mid-season, the change in ellipticity was helpful in separating crop types.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"793 - 813"},"PeriodicalIF":2.6,"publicationDate":"2022-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45005120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-15DOI: 10.1080/07038992.2022.2120858
K. Morrison, W. Wagner
Abstract In low-moisture regimes, strongly-reflecting bedrock underlying soil could provide a dominant return. This offers a novel opportunity to retrieve both the volumetric moisture fraction (mv ) and depth (d) of a soil layer using a differential phase. A radar wave traversing the overlying soil slows in response to moisture state; moisture dynamics are thus recorded as variations in travel time—captured back at a radar platform as changes in phase. The Phase Scaled Dielectric (PSD) model introduced here converts phase changes to those in soil dielectric as an intermediate step to estimating mv . Simulations utilizing a real soil moisture timeseries from a site in Sudan were used to demonstrate the linked behaviors of the soil and radar variables, and detail the PSD principle. A laboratory validation used soil with a wet top layer variable in depth 1–2 cm and drying from mv ∼ 0.2 m3m−3, overlying a gravel layer at a depth of 11 cm. The scheme retrieved = 1.49 ± 0.33 cm and a change Δmv = 0.191–0.021 ± 0.009 m3m−3. The PSD scheme outlined here promises a new avenue for the diagnostic measurement of soil parameters which is not currently available to radar remote sensing.
{"title":"Soil Moisture and Soil Depth Retrieval Using the Coupled Phase-Amplitude Behavior of C-Band Radar Backscatter in the Presence of Sub-Surface Scattering","authors":"K. Morrison, W. Wagner","doi":"10.1080/07038992.2022.2120858","DOIUrl":"https://doi.org/10.1080/07038992.2022.2120858","url":null,"abstract":"Abstract In low-moisture regimes, strongly-reflecting bedrock underlying soil could provide a dominant return. This offers a novel opportunity to retrieve both the volumetric moisture fraction (mv ) and depth (d) of a soil layer using a differential phase. A radar wave traversing the overlying soil slows in response to moisture state; moisture dynamics are thus recorded as variations in travel time—captured back at a radar platform as changes in phase. The Phase Scaled Dielectric (PSD) model introduced here converts phase changes to those in soil dielectric as an intermediate step to estimating mv . Simulations utilizing a real soil moisture timeseries from a site in Sudan were used to demonstrate the linked behaviors of the soil and radar variables, and detail the PSD principle. A laboratory validation used soil with a wet top layer variable in depth 1–2 cm and drying from mv ∼ 0.2 m3m−3, overlying a gravel layer at a depth of 11 cm. The scheme retrieved = 1.49 ± 0.33 cm and a change Δmv = 0.191–0.021 ± 0.009 m3m−3. The PSD scheme outlined here promises a new avenue for the diagnostic measurement of soil parameters which is not currently available to radar remote sensing.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"779 - 792"},"PeriodicalIF":2.6,"publicationDate":"2022-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48651120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-14DOI: 10.1080/07038992.2022.2118109
Yilong Lv, Min Li
Abstract Deep learning has been widely applied to ship detection in Synthetic Aperture Radar (SAR) images. Unlike optical images, the current object detection methods have the problem of weak feature representation due to the low object resolution in SAR images. In addition, disturbed by chaotic noise, the features of classification and location are prone to significant differences, resulting in classification and location task misalignment. Therefore, this paper proposes a novel SAR ship target detection algorithm based on Cross-Attention Mechanism (CAM), which can establish the information interaction between the classification and localization task and strengthen the correlation between features through attention. In addition, to suppress the noise in multi-scale feature fusion, we designed an Attention-based Feature Fusion Module (AFFM), which uses the attention information between channels to perform the re-weighting operation. This operation can enhance useful feature information and suppress noise information. Experimental results show that on a benchmark SAR Ship Detection Dataset (SSDD), the Fully Convolutional One-Stage Object Detector (FCOS) with ResNet-50 backbone network was optimized to improve AP by 6.5% and computational cost by 0.51%. RetinaNet with ResNet-50 backbone network was optimized to improve AP by 1.8% and computational cost by 0.51%.
{"title":"Ship Detection in SAR Images via Cross-Attention Mechanism","authors":"Yilong Lv, Min Li","doi":"10.1080/07038992.2022.2118109","DOIUrl":"https://doi.org/10.1080/07038992.2022.2118109","url":null,"abstract":"Abstract Deep learning has been widely applied to ship detection in Synthetic Aperture Radar (SAR) images. Unlike optical images, the current object detection methods have the problem of weak feature representation due to the low object resolution in SAR images. In addition, disturbed by chaotic noise, the features of classification and location are prone to significant differences, resulting in classification and location task misalignment. Therefore, this paper proposes a novel SAR ship target detection algorithm based on Cross-Attention Mechanism (CAM), which can establish the information interaction between the classification and localization task and strengthen the correlation between features through attention. In addition, to suppress the noise in multi-scale feature fusion, we designed an Attention-based Feature Fusion Module (AFFM), which uses the attention information between channels to perform the re-weighting operation. This operation can enhance useful feature information and suppress noise information. Experimental results show that on a benchmark SAR Ship Detection Dataset (SSDD), the Fully Convolutional One-Stage Object Detector (FCOS) with ResNet-50 backbone network was optimized to improve AP by 6.5% and computational cost by 0.51%. RetinaNet with ResNet-50 backbone network was optimized to improve AP by 1.8% and computational cost by 0.51%.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"764 - 778"},"PeriodicalIF":2.6,"publicationDate":"2022-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46025875","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-06DOI: 10.1080/07038992.2022.2110463
F. Ahern, B. Brisco, M. Battaglia, L. Bourgeau-Chavez, D. Atwood, K. Murnaghan
Abstract We have previously reported anomalous polarimetric decomposition results from SAR observations of wetlands. This is caused by the abrupt change in the phase difference between the HH and VV backscatter that occurs around the Brewster angle of the emergent vegetation. We have now developed and implemented a model for backscattering from wetlands that features a cylinder emergent from a water plane. The model was used in conjunction with an extensive set of RADARSAT-2 polarimetric observations of wetlands to provide further insights into the backscattering process. We are able to show how the abrupt Brewster transition in HH-VV phase difference varies with cylinder diameter and gravimetric moisture. We find that coherent cross-pol backscatter can result from cylindrical stems being tilted. In swamps with extensive tree mortality but primarily vertical trunks, the CPD can be used to monitor the drying of the trees and thus their fire hazard. These insights may be used to identify drying trees, indicating thawing permafrost, a potentially important climate change application in the near future. We recommend that applications researchers and users choose radar wavelengths that are considerably shorter, or longer, than the diameters of the cylinders producing the dominant double-bounce backscatter to avoid resonance effects.
{"title":"SAR Polarimetric Phase Differences in Wetlands: Information and Mis-Information","authors":"F. Ahern, B. Brisco, M. Battaglia, L. Bourgeau-Chavez, D. Atwood, K. Murnaghan","doi":"10.1080/07038992.2022.2110463","DOIUrl":"https://doi.org/10.1080/07038992.2022.2110463","url":null,"abstract":"Abstract We have previously reported anomalous polarimetric decomposition results from SAR observations of wetlands. This is caused by the abrupt change in the phase difference between the HH and VV backscatter that occurs around the Brewster angle of the emergent vegetation. We have now developed and implemented a model for backscattering from wetlands that features a cylinder emergent from a water plane. The model was used in conjunction with an extensive set of RADARSAT-2 polarimetric observations of wetlands to provide further insights into the backscattering process. We are able to show how the abrupt Brewster transition in HH-VV phase difference varies with cylinder diameter and gravimetric moisture. We find that coherent cross-pol backscatter can result from cylindrical stems being tilted. In swamps with extensive tree mortality but primarily vertical trunks, the CPD can be used to monitor the drying of the trees and thus their fire hazard. These insights may be used to identify drying trees, indicating thawing permafrost, a potentially important climate change application in the near future. We recommend that applications researchers and users choose radar wavelengths that are considerably shorter, or longer, than the diameters of the cylinders producing the dominant double-bounce backscatter to avoid resonance effects.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"703 - 721"},"PeriodicalIF":2.6,"publicationDate":"2022-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46920659","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-06DOI: 10.1080/07038992.2022.2117687
Caihong Gao, Qifan Wu, M. Dyck, Lei Fang, Hailong He
Abstract Greenhouses used for agricultural production have been expanding around the world because it significantly increases crop yield. Meanwhile, it brings a series of environmental problems that should be considered in agricultural planning and management. The advent of the Google Earth Engine (GEE) cloud platform makes remote sensing image processing more convenient and efficient. It has been widely applied in multiple disciplines, but few studies have investigated the detection of greenhouses. In this research, four different classification methods were applied for comparing their performance in monitoring greenhouses in the Guanzhong Plain, Shaanxi, China using GEE: the Minimum Distance Classifier (MDC), the Support Vector Machine with three kernel functions (linear, SVM-L, polynomial, SVM-P, and radial basis function variations, SVM-R), the Classification and Regression Trees (CART), and the Random Forest (RF). Our results illustrate that these classification techniques’ overall accuracy is >84%. The most accurate classification results were obtained by the SVM-R classifier, with an overall accuracy of 94%, followed by the RF and CART classifier, while the MDC performed worst among these four classifiers. These results would be useful for greenhouse extraction in long time series and large-scale areas, which provides solid information for decision-makers and practitioners for agriculture planning and management.
{"title":"Greenhouses Detection in Guanzhong Plain, Shaanxi, China: Evaluation of Four Classification Methods in Google Earth Engine","authors":"Caihong Gao, Qifan Wu, M. Dyck, Lei Fang, Hailong He","doi":"10.1080/07038992.2022.2117687","DOIUrl":"https://doi.org/10.1080/07038992.2022.2117687","url":null,"abstract":"Abstract Greenhouses used for agricultural production have been expanding around the world because it significantly increases crop yield. Meanwhile, it brings a series of environmental problems that should be considered in agricultural planning and management. The advent of the Google Earth Engine (GEE) cloud platform makes remote sensing image processing more convenient and efficient. It has been widely applied in multiple disciplines, but few studies have investigated the detection of greenhouses. In this research, four different classification methods were applied for comparing their performance in monitoring greenhouses in the Guanzhong Plain, Shaanxi, China using GEE: the Minimum Distance Classifier (MDC), the Support Vector Machine with three kernel functions (linear, SVM-L, polynomial, SVM-P, and radial basis function variations, SVM-R), the Classification and Regression Trees (CART), and the Random Forest (RF). Our results illustrate that these classification techniques’ overall accuracy is >84%. The most accurate classification results were obtained by the SVM-R classifier, with an overall accuracy of 94%, followed by the RF and CART classifier, while the MDC performed worst among these four classifiers. These results would be useful for greenhouse extraction in long time series and large-scale areas, which provides solid information for decision-makers and practitioners for agriculture planning and management.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"747 - 763"},"PeriodicalIF":2.6,"publicationDate":"2022-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43425429","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-05DOI: 10.1080/07038992.2022.2114439
Nicole Byford, C. Coburn
Abstract Spectral reflectance-based vicarious calibration (VicCal) requires accurate characterization of the bidirectional reflectance distribution function (BRDF) of the ground-based target. Railroad Valley (RRV) Playa, Nevada, has been used as a VicCal test site since 1995 as it is large, appears stable over time, and has a reasonably consistent surface. This study presents the results of a diurnal measurement cycle that closely replicated illumination geometries for Earth Observing (EO) satellites over a year. By measuring the rate of change of the BRDF with respect to time, we recorded the range of BRDF effects while holding the surface constant with respect to moisture and surface condition variation. The rate of spectral reflectance change increased rapidly with view angle in the backscatter direction, reaching rates of change that are 2.3 and 10.5 times greater in the backscatter than in the forward scatter direction for view angles of 20° and 40°, respectively. The results show that larger off-nadir viewing angles in the backscatter direction are particularly sensitive to changes in solar/view geometries.
{"title":"Temporal Variation in Surface Bidirectional Reflectance of the Railroad Valley Vicarious Calibration Test Site in Nevada","authors":"Nicole Byford, C. Coburn","doi":"10.1080/07038992.2022.2114439","DOIUrl":"https://doi.org/10.1080/07038992.2022.2114439","url":null,"abstract":"Abstract Spectral reflectance-based vicarious calibration (VicCal) requires accurate characterization of the bidirectional reflectance distribution function (BRDF) of the ground-based target. Railroad Valley (RRV) Playa, Nevada, has been used as a VicCal test site since 1995 as it is large, appears stable over time, and has a reasonably consistent surface. This study presents the results of a diurnal measurement cycle that closely replicated illumination geometries for Earth Observing (EO) satellites over a year. By measuring the rate of change of the BRDF with respect to time, we recorded the range of BRDF effects while holding the surface constant with respect to moisture and surface condition variation. The rate of spectral reflectance change increased rapidly with view angle in the backscatter direction, reaching rates of change that are 2.3 and 10.5 times greater in the backscatter than in the forward scatter direction for view angles of 20° and 40°, respectively. The results show that larger off-nadir viewing angles in the backscatter direction are particularly sensitive to changes in solar/view geometries.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"722 - 736"},"PeriodicalIF":2.6,"publicationDate":"2022-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46698611","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-03DOI: 10.1080/07038992.2022.2114440
Debaleena Datta, P. Mallick, Deepak Gupta, G. Chae
Abstract We propose a hybridized technique named Spatial-Spectral-Superpixelwise PCA-based Dense 2D-3D CNN Fusion Architecture (3SPCA-D-2D-3D-CNN), that deals with the detailed and complex study of dimensionality reduction and classification of Hyperspectal images (HSI). Our work is 2-fold: At first (1), 3SPCA is applied on raw HSI that adopts superpixels-based local reconstruction to filter the images, whereas PCA-based supplementary global features acquire the relevant and low-dimensional local features. Every HSI pixel is reconstituted by the pixels of its closest neighbors in the same superpixel block to reduce noise and improve spatial information. Next, PCA is conducted on every zone and the full HSI to get local and global features. The local-global and spatial-spectral properties are then concatenated. Secondly (2), the D-2D-3D-CNN fusion architecture is made up of three 3D convolution blocks, two 2D convolution blocks with varied kernel sizes and filters, and four fully connected (FC) dense layers, totaling nine distinguishing and information-enriched features. These features can generate precise class labels and apply them to the appropriate landcovers. The proposed method has been applied to three publicly available HSI landcover datasets, the Indian Pines, the Salinas Valley, and the Pavia University. It achieved respectively 98.33%, 99.99%, and 98.73% average accuracy scores. Due to its improved Feature Extraction capacity from a limited number of training samples and its classification performance with fewer epochs, this method outperforms other relevant state-of-the-art CNN-based methods.
{"title":"Hyperspectral Image Classification Based on Novel Hybridization of Spatial-Spectral-Superpixelwise Principal Component Analysis and Dense 2D-3D Convolutional Neural Network Fusion Architecture","authors":"Debaleena Datta, P. Mallick, Deepak Gupta, G. Chae","doi":"10.1080/07038992.2022.2114440","DOIUrl":"https://doi.org/10.1080/07038992.2022.2114440","url":null,"abstract":"Abstract We propose a hybridized technique named Spatial-Spectral-Superpixelwise PCA-based Dense 2D-3D CNN Fusion Architecture (3SPCA-D-2D-3D-CNN), that deals with the detailed and complex study of dimensionality reduction and classification of Hyperspectal images (HSI). Our work is 2-fold: At first (1), 3SPCA is applied on raw HSI that adopts superpixels-based local reconstruction to filter the images, whereas PCA-based supplementary global features acquire the relevant and low-dimensional local features. Every HSI pixel is reconstituted by the pixels of its closest neighbors in the same superpixel block to reduce noise and improve spatial information. Next, PCA is conducted on every zone and the full HSI to get local and global features. The local-global and spatial-spectral properties are then concatenated. Secondly (2), the D-2D-3D-CNN fusion architecture is made up of three 3D convolution blocks, two 2D convolution blocks with varied kernel sizes and filters, and four fully connected (FC) dense layers, totaling nine distinguishing and information-enriched features. These features can generate precise class labels and apply them to the appropriate landcovers. The proposed method has been applied to three publicly available HSI landcover datasets, the Indian Pines, the Salinas Valley, and the Pavia University. It achieved respectively 98.33%, 99.99%, and 98.73% average accuracy scores. Due to its improved Feature Extraction capacity from a limited number of training samples and its classification performance with fewer epochs, this method outperforms other relevant state-of-the-art CNN-based methods.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"663 - 680"},"PeriodicalIF":2.6,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42882039","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-03DOI: 10.1080/07038992.2022.2088486
Thiago Frank, A. Smith, B. Houston, Xiaoyu Yang, Xulin Guo
Abstract Estimating biophysical parameters of native grassland enables management changes that affect ecological processes and economic benefits. Although multiple hyperspectral studies were focused on native grasslands, just a few compare data at different scales and among ecoregions. In this study, we compared data collected at different spectral and spatial scales and among Canadian Prairie ecoregions. Field observations indicate that the Fescue Ecoregion grasslands has specific dominant species, while the Moist-Mixed and Mixed Ecoregions share similar dominant species, which is important in determining parameters such as leaf area index (LAI) and canopy height. Hyperspectral measurements showed a specific signature for the Fescue Ecoregion, due to denser canopies, while the Moist-Mixed and Mixed Ecoregions showed similar spectral characteristics to each other. The correlation between biophysical parameters and spectral indices reveals the importance of LAI, since it was significantly correlated with all spectral indices analyzed. The Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), and the Plant Senescence Reflectance Index (PSRI) showed significant correlations with biophysical parameters. The comparison results indicated the PSRI being overestimated at all sites (satellite data) and NDVI underestimated at all sites. Finally, the satellite-derived LAI showed a significant positive relationship with the field-measured LAI.
{"title":"Estimating Biophysical Parameters of Native Grasslands Using Spectral Data Derived from Close Range Hyperspectral and Satellite Data","authors":"Thiago Frank, A. Smith, B. Houston, Xiaoyu Yang, Xulin Guo","doi":"10.1080/07038992.2022.2088486","DOIUrl":"https://doi.org/10.1080/07038992.2022.2088486","url":null,"abstract":"Abstract Estimating biophysical parameters of native grassland enables management changes that affect ecological processes and economic benefits. Although multiple hyperspectral studies were focused on native grasslands, just a few compare data at different scales and among ecoregions. In this study, we compared data collected at different spectral and spatial scales and among Canadian Prairie ecoregions. Field observations indicate that the Fescue Ecoregion grasslands has specific dominant species, while the Moist-Mixed and Mixed Ecoregions share similar dominant species, which is important in determining parameters such as leaf area index (LAI) and canopy height. Hyperspectral measurements showed a specific signature for the Fescue Ecoregion, due to denser canopies, while the Moist-Mixed and Mixed Ecoregions showed similar spectral characteristics to each other. The correlation between biophysical parameters and spectral indices reveals the importance of LAI, since it was significantly correlated with all spectral indices analyzed. The Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), and the Plant Senescence Reflectance Index (PSRI) showed significant correlations with biophysical parameters. The comparison results indicated the PSRI being overestimated at all sites (satellite data) and NDVI underestimated at all sites. Finally, the satellite-derived LAI showed a significant positive relationship with the field-measured LAI.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"633 - 648"},"PeriodicalIF":2.6,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41428026","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：