Pub Date : 2022-09-03DOI: 10.1080/07038992.2022.2116566
M. A. Hamza, Jaber S. Alzahrani, Amal Al-Rasheed, R. Alshahrani, M. Alamgeer, Abdelwahed Motwakel, Ishfaq Yaseen, Mohamed I. Eldesouki
Abstract Unmanned Aerial Vehicle (UAV) is treated as an effective technique for gathering high resolution aerial images. The UAV based aerial image collection is highly preferred due to its inexpensive and effective nature. However, automatic classification of aerial images poses a major challenging issue in the design of UAV, which could be handled by the deep learning (DL) models. This study designs a novel UAV assisted DL based image classification model (UAVDL-ICM) for Industry 4.0 environment. The proposed UAVDL-ICM technique involves an ensemble of voting based three DL models, namely Residual network (ResNet), Inception with ResNetv2, and Densely Connected Networks (DenseNet). Also, the hyperparameter tuning of these DL models takes place using a genetic programming (GP) approach. Finally, Oppositional Water Wave Optimization (OWWO) with Fully Connected Deep Neural networks (FCDNN) is employed for the classification of aerial images. A wide range of simulations takes place and the results are examined in terms of different parameters. A detailed comparative study highlighted the betterment of the UAVDL-ICM technique compared to other recent approaches.
{"title":"Optimal and Fully Connected Deep Neural Networks Based Classification Model for Unmanned Aerial Vehicle Using Hyperspectral Remote Sensing Images","authors":"M. A. Hamza, Jaber S. Alzahrani, Amal Al-Rasheed, R. Alshahrani, M. Alamgeer, Abdelwahed Motwakel, Ishfaq Yaseen, Mohamed I. Eldesouki","doi":"10.1080/07038992.2022.2116566","DOIUrl":"https://doi.org/10.1080/07038992.2022.2116566","url":null,"abstract":"Abstract Unmanned Aerial Vehicle (UAV) is treated as an effective technique for gathering high resolution aerial images. The UAV based aerial image collection is highly preferred due to its inexpensive and effective nature. However, automatic classification of aerial images poses a major challenging issue in the design of UAV, which could be handled by the deep learning (DL) models. This study designs a novel UAV assisted DL based image classification model (UAVDL-ICM) for Industry 4.0 environment. The proposed UAVDL-ICM technique involves an ensemble of voting based three DL models, namely Residual network (ResNet), Inception with ResNetv2, and Densely Connected Networks (DenseNet). Also, the hyperparameter tuning of these DL models takes place using a genetic programming (GP) approach. Finally, Oppositional Water Wave Optimization (OWWO) with Fully Connected Deep Neural networks (FCDNN) is employed for the classification of aerial images. A wide range of simulations takes place and the results are examined in terms of different parameters. A detailed comparative study highlighted the betterment of the UAVDL-ICM technique compared to other recent approaches.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"681 - 693"},"PeriodicalIF":2.6,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42925727","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: 10.1080/07038992.2022.2123625
Gautam Srivastava, K. Shankar
Hyperspectral Remote Sensing (HRS) is an emerging, multidisciplinary paradigm with a variety of applications that are built on the principle of material spectroscopy, radiative transfer, and imaging spectroscopy. HRS acquires digital imagery of materials in many narrow contiguous spectral bands. HRS provides high spatial/spectral resolution data for each picture element (pixel). Targets are identified based on the physical information extracted from the spectrum and are quantitatively analyzed in the spatial view. The most crucial and efficient advantage of HRS is that it can acquire quantitative information from many points on the ground at the same instant of time. Regarding this multidisciplinary paradigm, HRS has several applications that lead to improvements in our digital lives. Utilizing HRS for monitoring and mapping changes in different areas around Earth will play an extensive and significant role in oceanography, agriculture, atmosphere, geology, hydrology, etc. In oceanography, it helps to classify and quantify complex oceanic environments and it also develops optically based chemical sensors for monitoring ecologically important nutrients and potentially harmful pollutants. Moreover, the high spectral resolution of HRS has an extra intelligence of capturing and discriminating subtle differences among crop types and also advancement in understanding the changes in biochemical and biophysical attributes of the crops. Furthermore, in the area of Hydrology, HRS has been used for monitoring water quality conditions of open water aquatic ecosystems and also identifies various water quality parameters like temperature, chlorophyll phosphorus, and turbidity. Considering smart technologies, HRS facilitates the characterization, and mapping of soil on a regional scale which includes soil mixture monitoring, weather monitoring, and atmospheric monitoring. Like all other existing remote sensing systems, HRS also faces some challenges in the optimal utilization of systems in various areas. The crucial factors to be considered while designing HRS for Earth monitoring and mapping is that it requires professional manpower to operate, and process the data while also requiring huge implementation costs. To overcome these challenges, various types of research are explored to investigate the implementation of HRS for EDITORIAL
{"title":"Advances in Hyperspectral Remote Sensing for Earth Monitoring and Mapping","authors":"Gautam Srivastava, K. Shankar","doi":"10.1080/07038992.2022.2123625","DOIUrl":"https://doi.org/10.1080/07038992.2022.2123625","url":null,"abstract":"Hyperspectral Remote Sensing (HRS) is an emerging, multidisciplinary paradigm with a variety of applications that are built on the principle of material spectroscopy, radiative transfer, and imaging spectroscopy. HRS acquires digital imagery of materials in many narrow contiguous spectral bands. HRS provides high spatial/spectral resolution data for each picture element (pixel). Targets are identified based on the physical information extracted from the spectrum and are quantitatively analyzed in the spatial view. The most crucial and efficient advantage of HRS is that it can acquire quantitative information from many points on the ground at the same instant of time. Regarding this multidisciplinary paradigm, HRS has several applications that lead to improvements in our digital lives. Utilizing HRS for monitoring and mapping changes in different areas around Earth will play an extensive and significant role in oceanography, agriculture, atmosphere, geology, hydrology, etc. In oceanography, it helps to classify and quantify complex oceanic environments and it also develops optically based chemical sensors for monitoring ecologically important nutrients and potentially harmful pollutants. Moreover, the high spectral resolution of HRS has an extra intelligence of capturing and discriminating subtle differences among crop types and also advancement in understanding the changes in biochemical and biophysical attributes of the crops. Furthermore, in the area of Hydrology, HRS has been used for monitoring water quality conditions of open water aquatic ecosystems and also identifies various water quality parameters like temperature, chlorophyll phosphorus, and turbidity. Considering smart technologies, HRS facilitates the characterization, and mapping of soil on a regional scale which includes soil mixture monitoring, weather monitoring, and atmospheric monitoring. Like all other existing remote sensing systems, HRS also faces some challenges in the optimal utilization of systems in various areas. The crucial factors to be considered while designing HRS for Earth monitoring and mapping is that it requires professional manpower to operate, and process the data while also requiring huge implementation costs. To overcome these challenges, various types of research are explored to investigate the implementation of HRS for EDITORIAL","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"575 - 578"},"PeriodicalIF":2.6,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42932926","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: 10.1080/07038992.2022.2116307
Guangyi Chen, A. Krzyżak, S. Qian
Abstract Hyperspectral imagery (HSI) denoising is a popular research topic in remote sensing. In this paper, we propose a novel method for HSI denoising by performing Minimum Noise Fraction (MNF) to the original HSI data cube, thresholding the noisy output bands with the Video Non-Local Bayes (VNLB) algorithm, and then conducting the inverse MNF transform to obtain the denoised data cube. Our experiments demonstrate that the proposed method usually achieves the best denoising results among several existing denoising methods for two HSI data cubes. In addition, it is much faster for HSI denoising than the VNLB algorithm which was originally developed for video denoising.
{"title":"Hyperspectral Imagery Denoising Using Minimum Noise Fraction and Video Non-Local Bayes Algorithms","authors":"Guangyi Chen, A. Krzyżak, S. Qian","doi":"10.1080/07038992.2022.2116307","DOIUrl":"https://doi.org/10.1080/07038992.2022.2116307","url":null,"abstract":"Abstract Hyperspectral imagery (HSI) denoising is a popular research topic in remote sensing. In this paper, we propose a novel method for HSI denoising by performing Minimum Noise Fraction (MNF) to the original HSI data cube, thresholding the noisy output bands with the Video Non-Local Bayes (VNLB) algorithm, and then conducting the inverse MNF transform to obtain the denoised data cube. Our experiments demonstrate that the proposed method usually achieves the best denoising results among several existing denoising methods for two HSI data cubes. In addition, it is much faster for HSI denoising than the VNLB algorithm which was originally developed for video denoising.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"694 - 701"},"PeriodicalIF":2.6,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42011443","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 : 2022-08-25DOI: 10.1080/07038992.2022.2114891
Hongzhang Ma, Shuyi Sun, Zhaowei Wang, Yandi Jiang, Sumei Liu
Abstract Soil Moisture (SM) plays a key role in the energy exchange between the atmosphere and the land surface. Most of the SM products retrieved from satellite remote sensing data are not suitable for drought monitoring and irrigation management in smart agriculture applications due to their coarse spatial resolution. We propose an SM downscaling method named Water Cloud Change Detection (WCCD) that effectively combines the Water-Cloud Model (WCM) and the Change Detection Method (CDM) to downscale the China Land Data Assimilation System soil moisture (CLDAS_SM, 6000-m resolution) product. The WCM is used to retrieve the soil backscattering at a fine spatial resolution by deducting the canopy backscattering from the surface total backscattering, and the linear regression relationship between soil backscattering and CLDAS_SM is established for each pixel at the coarse scale under the assumption that the surface roughness does not change for dozens of days. The performance of the algorithm is tested in an agricultural crop region in Hebei province of China with Sentinel-1 and Sentinel-2 images. The validation results show that the downscaled SM at different spatial resolutions are in good agreement with the in-situ measurements with the correlation coefficient (R) higher than 0.71 and the Root Mean Squared Error (RMSE) lower than 0.042 cm3×cm−3.
{"title":"Downscaling CLDAS Soil Moisture Product by Integrating Sentinel-1 and Sentinel-2 Data over Agricultural Area","authors":"Hongzhang Ma, Shuyi Sun, Zhaowei Wang, Yandi Jiang, Sumei Liu","doi":"10.1080/07038992.2022.2114891","DOIUrl":"https://doi.org/10.1080/07038992.2022.2114891","url":null,"abstract":"Abstract Soil Moisture (SM) plays a key role in the energy exchange between the atmosphere and the land surface. Most of the SM products retrieved from satellite remote sensing data are not suitable for drought monitoring and irrigation management in smart agriculture applications due to their coarse spatial resolution. We propose an SM downscaling method named Water Cloud Change Detection (WCCD) that effectively combines the Water-Cloud Model (WCM) and the Change Detection Method (CDM) to downscale the China Land Data Assimilation System soil moisture (CLDAS_SM, 6000-m resolution) product. The WCM is used to retrieve the soil backscattering at a fine spatial resolution by deducting the canopy backscattering from the surface total backscattering, and the linear regression relationship between soil backscattering and CLDAS_SM is established for each pixel at the coarse scale under the assumption that the surface roughness does not change for dozens of days. The performance of the algorithm is tested in an agricultural crop region in Hebei province of China with Sentinel-1 and Sentinel-2 images. The validation results show that the downscaled SM at different spatial resolutions are in good agreement with the in-situ measurements with the correlation coefficient (R) higher than 0.71 and the Root Mean Squared Error (RMSE) lower than 0.042 cm3×cm−3.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"737 - 746"},"PeriodicalIF":2.6,"publicationDate":"2022-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41721809","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 : 2022-07-04DOI: 10.1080/07038992.2022.2096579
Na Wang, Z. Shi, Zhaoxu Zhang
Abstract According to the spatial structure characteristics of road curbs and road surfaces, a robust method for automatic extraction of road boundaries, road curbs and road surfaces was proposed using mobile laser scanning (MLS) point cloud data. Firstly, ground filtering was performed to separate ground points and non-ground points according to the angle between the normal vector of the point cloud and the direction vector of the z-axis. Secondly, based on the vertical and linear features of the road curb, the MLS trajectory points were used to extract road curb and road boundary points. Then, Euclidean clustering and fitting were performed on the road boundary point segments. Adjacent clusters were merged, and sparse points were densified. In addition, based on the principle that road surfaces are within road boundaries, road surface points were obtained in scanning line order. Two MLS point clouds with different resolutions and road roughness were tested. Compared with the manually calibrated reference road curb, the extraction completenesses of the road curb from the two datasets were 95.66% and 96.45%, respectively, and the extraction correctnesses of the road curb were 96.34% and 99.10%, respectively, with both qualities over 92%. The algorithm can effectively extract straight or curved road boundaries and road curbs from the point cloud data containing vehicles, pedestrians and obstacle occlusions in an urban environment, and is applicable to MLS point cloud data with different resolutions and roughness.
{"title":"Road Boundary, Curb and Surface Extraction from 3D Mobile LiDAR Point Clouds in Urban Environment","authors":"Na Wang, Z. Shi, Zhaoxu Zhang","doi":"10.1080/07038992.2022.2096579","DOIUrl":"https://doi.org/10.1080/07038992.2022.2096579","url":null,"abstract":"Abstract According to the spatial structure characteristics of road curbs and road surfaces, a robust method for automatic extraction of road boundaries, road curbs and road surfaces was proposed using mobile laser scanning (MLS) point cloud data. Firstly, ground filtering was performed to separate ground points and non-ground points according to the angle between the normal vector of the point cloud and the direction vector of the z-axis. Secondly, based on the vertical and linear features of the road curb, the MLS trajectory points were used to extract road curb and road boundary points. Then, Euclidean clustering and fitting were performed on the road boundary point segments. Adjacent clusters were merged, and sparse points were densified. In addition, based on the principle that road surfaces are within road boundaries, road surface points were obtained in scanning line order. Two MLS point clouds with different resolutions and road roughness were tested. Compared with the manually calibrated reference road curb, the extraction completenesses of the road curb from the two datasets were 95.66% and 96.45%, respectively, and the extraction correctnesses of the road curb were 96.34% and 99.10%, respectively, with both qualities over 92%. The algorithm can effectively extract straight or curved road boundaries and road curbs from the point cloud data containing vehicles, pedestrians and obstacle occlusions in an urban environment, and is applicable to MLS point cloud data with different resolutions and roughness.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"504 - 519"},"PeriodicalIF":2.6,"publicationDate":"2022-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49661202","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 : 2022-07-04DOI: 10.1080/07038992.2022.2096580
Savannah Bussières, C. Kinnard, Maxime Clermont, Stéphane Campeau, Daphney Dubé-Richard, Pierre-André Bordeleau, Alexandre Roy
Abstract The Lake Saint-Pierre (LSP) is a wide (≈300 km2) and shallow (≈3 m) lake created through a widening of the St. Lawrence River. Each spring, freshet makes it the largest floodplain in the province of Quebec. Agricultural practices in the littoral increase the water turbidity, which deteriorate the habitat’s quality of many fish species. However, measuring spatio-temporal turbidity patterns in the LSP floodplain remain difficult because turbidity is highly variable in space and time. This study aims to evaluate the potential to use an Unmanned Aerial Vehicle (UAV) to measure the water turbidity in the LSP’s floodplain. The results show that the UAV can efficiently measure the variation of turbidity in the LSP with a RMSE of 28.22 FNU. We also compared the turbidity retrieved from UAV with those retrieved from Sentinel-2 observations. The results show that the two models are comparable, even if Sentinel-2 yields better results. However, challenges remain when using UAV for turbidity monitoring, such as software limitations for mosaics creation over large water bodies. Nevertheless, the high spatial and temporal information can provide insights into the complex water turbidity patterns which characterize floodplains. The method could help land use management to improve the water quality of these ecosystems.
{"title":"Monitoring Water Turbidity in a Temperate Floodplain Using UAV: Potential and Challenges","authors":"Savannah Bussières, C. Kinnard, Maxime Clermont, Stéphane Campeau, Daphney Dubé-Richard, Pierre-André Bordeleau, Alexandre Roy","doi":"10.1080/07038992.2022.2096580","DOIUrl":"https://doi.org/10.1080/07038992.2022.2096580","url":null,"abstract":"Abstract The Lake Saint-Pierre (LSP) is a wide (≈300 km2) and shallow (≈3 m) lake created through a widening of the St. Lawrence River. Each spring, freshet makes it the largest floodplain in the province of Quebec. Agricultural practices in the littoral increase the water turbidity, which deteriorate the habitat’s quality of many fish species. However, measuring spatio-temporal turbidity patterns in the LSP floodplain remain difficult because turbidity is highly variable in space and time. This study aims to evaluate the potential to use an Unmanned Aerial Vehicle (UAV) to measure the water turbidity in the LSP’s floodplain. The results show that the UAV can efficiently measure the variation of turbidity in the LSP with a RMSE of 28.22 FNU. We also compared the turbidity retrieved from UAV with those retrieved from Sentinel-2 observations. The results show that the two models are comparable, even if Sentinel-2 yields better results. However, challenges remain when using UAV for turbidity monitoring, such as software limitations for mosaics creation over large water bodies. Nevertheless, the high spatial and temporal information can provide insights into the complex water turbidity patterns which characterize floodplains. The method could help land use management to improve the water quality of these ecosystems.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"565 - 574"},"PeriodicalIF":2.6,"publicationDate":"2022-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46184297","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}
Abstract At present, most machine learning bathymetry retrieval models use the band reflectance as the inversion feature only, without considering features related to the water substrate and pixel spatial correlation. In this study, in addition to band reflectance, two features, Depth-Invariant Index (DII) and pixel location, are taken into account. Two machine learning algorithms, Random Forest (RF) and Back Propagation (BP) neural network are used to retrieve bathymetry. The effects of the two features on the accuracy and performance of bathymetry retrieval are explored. The results show that: (i) Machine learning algorithms are generally superior to the widely used Stumpf model. Stumpf model performs better only in the depth range of 8–16 m, with a Root Mean Square Error (RMSE) of 0.85 m, but has poor performance in other depth ranges. (ii) Compared with models that use Band Reflectance (BR) only, DIIb,g (blue-green DII) + BR model, Location and Location + BR models are all superior to the BR model for RF and BP algorithms. It means that DII and location features are very effective in improving the bathymetry retrieval accuracy.
{"title":"Shallow Water Bathymetry Retrieval by Optical Remote Sensing Based on Depth-Invariant Index and Location Features","authors":"Jinshan Zhu, Fei Yin, Jian Qin, Jiawei Qi, Zhaoyu Ren, Peng Hu, Jingyu Zhang, Xueqing Zhang, Ruifu Wang","doi":"10.1080/07038992.2022.2104235","DOIUrl":"https://doi.org/10.1080/07038992.2022.2104235","url":null,"abstract":"Abstract At present, most machine learning bathymetry retrieval models use the band reflectance as the inversion feature only, without considering features related to the water substrate and pixel spatial correlation. In this study, in addition to band reflectance, two features, Depth-Invariant Index (DII) and pixel location, are taken into account. Two machine learning algorithms, Random Forest (RF) and Back Propagation (BP) neural network are used to retrieve bathymetry. The effects of the two features on the accuracy and performance of bathymetry retrieval are explored. The results show that: (i) Machine learning algorithms are generally superior to the widely used Stumpf model. Stumpf model performs better only in the depth range of 8–16 m, with a Root Mean Square Error (RMSE) of 0.85 m, but has poor performance in other depth ranges. (ii) Compared with models that use Band Reflectance (BR) only, DIIb,g (blue-green DII) + BR model, Location and Location + BR models are all superior to the BR model for RF and BP algorithms. It means that DII and location features are very effective in improving the bathymetry retrieval accuracy.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"534 - 550"},"PeriodicalIF":2.6,"publicationDate":"2022-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48714458","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 : 2022-07-04DOI: 10.1080/07038992.2022.2103397
Jean Xiong, Ting Chen, Minjie Wang, Jianjun He, Lanying Wang, Zhiyong Wang
Abstract Automatically mapping building footprints has a wide range of applications in many fields. In recent years, the automatic building extraction methods based on deep learning show an absolute advantage over the traditional image segmentation methods due to its high speed and high precision. However, the building footprint extracted by deep learning is just an irregular building mask. There is still much work to be done to transform the building mask into a vector building footprint in the usual sense. One of the most important tasks is to determine the orientation of each side of the building. Most of the current methods are based on the building mask to determine the orientation of each side of the building. The biggest disadvantage of this method is that it completely relies on the building mask which is often unsatisfactory. In this case, the article proposes a method to determine the orientation of each side of the building based on the building mask and line segments, thereby effectively avoiding the danger of relying on the building mask. Experiments show that the proposed method can achieve high-speed and high-precision automatic extraction of building footprints from remote sensing images, saving costs.
{"title":"A Method for Fully Automatic Building Footprint Extraction From Remote Sensing Images","authors":"Jean Xiong, Ting Chen, Minjie Wang, Jianjun He, Lanying Wang, Zhiyong Wang","doi":"10.1080/07038992.2022.2103397","DOIUrl":"https://doi.org/10.1080/07038992.2022.2103397","url":null,"abstract":"Abstract Automatically mapping building footprints has a wide range of applications in many fields. In recent years, the automatic building extraction methods based on deep learning show an absolute advantage over the traditional image segmentation methods due to its high speed and high precision. However, the building footprint extracted by deep learning is just an irregular building mask. There is still much work to be done to transform the building mask into a vector building footprint in the usual sense. One of the most important tasks is to determine the orientation of each side of the building. Most of the current methods are based on the building mask to determine the orientation of each side of the building. The biggest disadvantage of this method is that it completely relies on the building mask which is often unsatisfactory. In this case, the article proposes a method to determine the orientation of each side of the building based on the building mask and line segments, thereby effectively avoiding the danger of relying on the building mask. Experiments show that the proposed method can achieve high-speed and high-precision automatic extraction of building footprints from remote sensing images, saving costs.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"520 - 533"},"PeriodicalIF":2.6,"publicationDate":"2022-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45961579","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 : 2022-06-27DOI: 10.1080/07038992.2022.2089102
A. Dutta, Majed Alsanea, B. Qureshi, Faisal Yousef Alghayadh, A. R. W. Sait
Abstract Hyperspectral imaging (HSI) can be attained by the use of high resolution optical sensors and it comprises several spectral bands of the identical remote sensing target and is treated as a three-dimension (3D) dataset. Recently, deep learning (DL) techniques are gained important attention among research communities for image classification. In this aspect, this study develops an intelligent rider optimization algorithm with deep learning enabled HSI classification model, named IRODL-HSIC technique. The proposed IRODL-HSIC technique aims to categorize the different class labels of the multispectral images. Besides, the IRODL-HSIC technique applies singular value decomposition. Moreover, the ResNet-152 technique was executed as a feature extractor to generate a collection of features. Furthermore, the rider optimization algorithm with cascaded recurrent neural network (CRNN) approach is utilized for the classification process. For ensuring the enhanced performance of the IRODL-HSIC algorithm, a wide range of simulations take place utilizing the multispectral images and the outcomes are examined under different aspects. The extensive comparative study highlighted the better performance of the IRODL-HSIC technique over the recent methods.
{"title":"Intelligent Rider Optimization Algorithm with Deep Learning Enabled Hyperspectral Remote Sensing Imaging Classification","authors":"A. Dutta, Majed Alsanea, B. Qureshi, Faisal Yousef Alghayadh, A. R. W. Sait","doi":"10.1080/07038992.2022.2089102","DOIUrl":"https://doi.org/10.1080/07038992.2022.2089102","url":null,"abstract":"Abstract Hyperspectral imaging (HSI) can be attained by the use of high resolution optical sensors and it comprises several spectral bands of the identical remote sensing target and is treated as a three-dimension (3D) dataset. Recently, deep learning (DL) techniques are gained important attention among research communities for image classification. In this aspect, this study develops an intelligent rider optimization algorithm with deep learning enabled HSI classification model, named IRODL-HSIC technique. The proposed IRODL-HSIC technique aims to categorize the different class labels of the multispectral images. Besides, the IRODL-HSIC technique applies singular value decomposition. Moreover, the ResNet-152 technique was executed as a feature extractor to generate a collection of features. Furthermore, the rider optimization algorithm with cascaded recurrent neural network (CRNN) approach is utilized for the classification process. For ensuring the enhanced performance of the IRODL-HSIC algorithm, a wide range of simulations take place utilizing the multispectral images and the outcomes are examined under different aspects. The extensive comparative study highlighted the better performance of the IRODL-HSIC technique over the recent methods.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"649 - 662"},"PeriodicalIF":2.6,"publicationDate":"2022-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48859365","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 : 2022-06-22DOI: 10.1080/07038992.2022.2081538
José Escorcia-Gutierrez, Margarita Gamarra, Melitsa Torres-Torres, Natasha Madera, Juan C. Calabria-Sarmiento, R. Mansour
Abstract Hyperspectral Remote Sensing (HRS) is an emergent, multidisciplinary paradigm with several applications, which are developed on the basis of material spectroscopy, radiative transfer, and imaging spectroscopy. HRS plays a vital role in agriculture for crops type classification and soil prediction. The recently developed artificial intelligence techniques can be used for crops type classification using HRS. This study develops an Intelligent Sine Cosine Optimization with Deep Transfer Learning Based Crop Type Classification (ISCO-DTLCTC) model. The ISCO-DTLCTC technique comprises initial preprocessing step to extract the region of interest. The information gain-based feature reduction technique is employed to reduce the dimensionality of the original hyperspectral images. In addition, a fusion of 3 deep convolutional neural networks models namely, VGG16, SqueezeNet, and Dense-EfficientNet perform feature extraction process. Furthermore, sine cosine optimization (SCO) algorithm with Modified Elman Neural Network (MENN) model is applied for crops type classification. The design of SCO algorithm helps to proficiently select the parameters involved in the MENN model. The performance validation of the ISCO-DTLCTC model is carried out using benchmark datasets and the results are inspected under several measures. Extensive comparative results demonstrated the betterment of the ISCO-DTLCTC model over the state of art approaches with maximum accuracy of 99.99%.
{"title":"Intelligent Sine Cosine Optimization with Deep Transfer Learning Based Crops Type Classification Using Hyperspectral Images","authors":"José Escorcia-Gutierrez, Margarita Gamarra, Melitsa Torres-Torres, Natasha Madera, Juan C. Calabria-Sarmiento, R. Mansour","doi":"10.1080/07038992.2022.2081538","DOIUrl":"https://doi.org/10.1080/07038992.2022.2081538","url":null,"abstract":"Abstract Hyperspectral Remote Sensing (HRS) is an emergent, multidisciplinary paradigm with several applications, which are developed on the basis of material spectroscopy, radiative transfer, and imaging spectroscopy. HRS plays a vital role in agriculture for crops type classification and soil prediction. The recently developed artificial intelligence techniques can be used for crops type classification using HRS. This study develops an Intelligent Sine Cosine Optimization with Deep Transfer Learning Based Crop Type Classification (ISCO-DTLCTC) model. The ISCO-DTLCTC technique comprises initial preprocessing step to extract the region of interest. The information gain-based feature reduction technique is employed to reduce the dimensionality of the original hyperspectral images. In addition, a fusion of 3 deep convolutional neural networks models namely, VGG16, SqueezeNet, and Dense-EfficientNet perform feature extraction process. Furthermore, sine cosine optimization (SCO) algorithm with Modified Elman Neural Network (MENN) model is applied for crops type classification. The design of SCO algorithm helps to proficiently select the parameters involved in the MENN model. The performance validation of the ISCO-DTLCTC model is carried out using benchmark datasets and the results are inspected under several measures. Extensive comparative results demonstrated the betterment of the ISCO-DTLCTC model over the state of art approaches with maximum accuracy of 99.99%.","PeriodicalId":48843,"journal":{"name":"Canadian Journal of Remote Sensing","volume":"48 1","pages":"621 - 632"},"PeriodicalIF":2.6,"publicationDate":"2022-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42419610","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：