Iman Khosravi | University of Isfahan (original) (raw)

Papers by Iman Khosravi

Research paper thumbnail of Presenting an extended evaluation framework for building detection algorithms using high spatial resolution images

This paper aims to provide an extended evaluation framework for building detection algorithms usi... more This paper aims to provide an extended evaluation framework for building detection algorithms using a diverse set of High Spatial Resolution (HSR) images. The HSR images utilized in this paper were chosen from different places and different sensors, and based on several important challenges in an urban area such as building alignment, density, shape, size, color, height, and imaging angle. The classical evaluation metrics such as detection rate, reliability, false positive rate, and overall accuracy only demonstrate the performance evaluation of an algorithm in relation to the buildings and cannot interpret the mentioned challenges. The extended evaluation framework proposed in this paper composed several extended metrics for performance evaluation of building detection algorithms in relation to these challenges in addition to the classical metrics. The paper intends to declare that the success or failure metrics of a building detection algorithm can have more varieties. In fact, a ...

Research paper thumbnail of Fully polarimetric synthetic aperture radar data classification using probabilistic and non-probabilistic kernel methods

European Journal of Remote Sensing, 2021

The data classification of fully polarimetric synthetic aperture radar (PolSAR) is one of the fav... more The data classification of fully polarimetric synthetic aperture radar (PolSAR) is one of the favourite topics in the remote sensing community. To date, a wide variety of algorithms have been utilized for PolSAR data classification, and among them kernel methods are the most attractive algorithms for this purpose. The most famous kernel method, i.e., the support vector machine (SVM) has been widely used for PolSAR data classification. However, until now, no studies to classify PolSAR data have been carried out using certain extended SVM versions, such as the least squares support vector machine (LSSVM), relevance vector machine (RVM) and import vector machine (IVM). Therefore, this work has employed and compared these four kernel methods for the classification of three PolSAR data sets. These methods were compared in two groups: the SVM and LSSVM as non-probabilistic kernel methods vs. the RVM and IVM as probabilistic kernel methods. In general, the results demonstrated that the SVM was marginally better, more accurate and more stable than the probabilistic kernels. Furthermore, the LSSVM performed much faster than the probabilistic kernel methods and its associated version, the SVM, with comparable accuracy.

Research paper thumbnail of The Use of Multiple Classifier Systems For Improving the Classification Accuracy of High–Dimensional Fully Polarimetric SAR Images

Journal of Geospatial Information Technology, 2014

A fully polarimetric synthetic aperture radar (POLSAR) image can provide a high-dimensional data.... more A fully polarimetric synthetic aperture radar (POLSAR) image can provide a high-dimensional data. This large amount of information can increase the overall accuracy of land-cover classification. But increasing the data dimensions if inadequately number of training samples may increase the complexity and cause the curse of dimensionality phenomenon. One of the strategies for solving this problem is the use of multiple classifier systems (MCS) that has the capability of divide and conquer to the large data as compared to the individual classifiers. In addition, some of MCS methods using the weak and unstable classifiers such as decision tree (DT) and neural network (NN) can obtain the high accuracy in high-dimensional data. The objective of this paper is also to use several popular MCS methods such as adaboost, bagging and random forests in order to improve the accuracy of land-cover classification from high-dimensional PolSAR images. The data used in this paper are Radarsat-2 image from San Francisco Bay and AIRSAR image of Flevoland. For classifying two these images, 69 polarimetric features were extracted from them. Two classifiers of DT and NN were chosen as the base classifiers of adaboost and bagging methods. In the next, the MCS methods were compared with the individual classifiers of DT and NN. The results indicated the higher overall accuracy of MCS methods between 5%-8% for classifying first image and 9%-16% for classifying second image. Even, the producer's accuracy and user's accuracy of MCS methods at all classes were more than the those of individual classifiers. So that at some classes, the difference was between 20% to even near 50%. These results confirmed that the MCS methods not only can produce higher overall accuracy at land-cover classification, but also they have the higher efficiency and reliability at discriminate individual classes.

Research paper thumbnail of Presenting A Feature Selection Method Based On Genetic Algorithm and Decision Tree For Classifying Fully Polarimetric SAR Images

Journal of Geospatial Information Technology, 2015

A fully polarimetric synthetic aperture radar (POLSAR) image can provide important polarimetric f... more A fully polarimetric synthetic aperture radar (POLSAR) image can provide important polarimetric features for land cover classification. These features can be the parameters obtained from scatering, covariance and coherency matrices, parameters extracted from target decomposition methods or both of them. In this paper, many polarimetric features are extracted from a POLSAR image. Then, with the use of Genetic Algorithm (GA) and Decision Tree (DT), a feature selection method based on the classification is presented. Afterwards, a comparative analysis is accomplished between DT classification with features selected from the proposed method and DT classification with all features. Moreover, the proposed method should be compared with the feature selection method of GA and Support Vector Machine (SVM). The results indicated that the accuracy of the proposed method (DT classification with the features selected from GA-DT algorithm) is nearly 3% higher than the ones of the DT classification with all features and it is approximately equal with the ones of the DT classification with the features selected from GA-SVM algorithm. However, the performance speed of the proposed method is approximately 5 times more than the ones of DT classification with the features selected from GA-SVM algorithm. As an another result, the features selected from the proposed method have a more success than the ones of two other methods at classifying the urban areas and vegetation classes.

Research paper thumbnail of MSMD: maximum separability and minimum dependency feature selection for cropland classification from optical and radar data

International Journal of Remote Sensing, 2018

ABSTRACT Cropland classification using optical and full polarimetric synthetic aperture radar (Po... more ABSTRACT Cropland classification using optical and full polarimetric synthetic aperture radar (PolSAR) images is a topic of considerable interest in the remote-sensing community. These two data sources can provide a diverse set of temporal, spectral, textural and polarimetric features which can be invaluable for cropland classification. However, some optical features or some radar features may have a relatively high correlation with other features. Hence, it seems to be necessary to choose the optimum features in order to reduce the dimensions of the data and to improve cropland classification accuracy. This article proposes a strategic feature selection method from a feature set of bitemporal RapidEye and Uninhabited Aerial Vehicle synthetic aperture radar (UAVSAR) images. The proposed method is designed to select the most relevant features and to remove redundant features based on the two concepts of separability and dependency. The proposed method is therefore referred to as maximum separability and minimum dependency (MSMD). For evaluating efficiency, MSMD and some well-known filter and wrapper feature selection methods are compared using a random forest classifier. Experimental tests confirmed that the classification results obtained from the MSMD feature selection method were more accurate than those achieved by filter methods. Moreover, they had an accuracy comparable to that of the results from the wrapper method. Furthermore, with regard to running time, MSMD operated as fast as the filter methods. It had a straightforward structure compared to the wrapper method, and as a result was faster than this method.

Research paper thumbnail of A random forest-based framework for crop mapping using temporal, spectral, textural and polarimetric observations

International Journal of Remote Sensing, 2019

Combining optical and polarimetric synthetic aperture radar (PolSAR) earth observations offers a ... more Combining optical and polarimetric synthetic aperture radar (PolSAR) earth observations offers a complementary data set with a significant number of spectral, textural, and polarimetric features for crop mapping and monitoring. Moreover, a temporal combination of both sources of information may lead to obtaining more reliable results compared to the use of single-time observations. In this paper, an operational framework based on the stacked generalization of random forest (RF), which efficiently employed bi-temporal observations of optical and radar data, was proposed for crop mapping. In the first step, various spectral, vegetation index, textural, and polarimetric features were extracted from both data sources and placed into several groups. Each group was classified separately using a single RF classifier. Then, several additional classification tasks were accomplished by another RF classifier. The earth observations used in this paper were collected by RapidEye satellites and the Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) system over an agricultural region near Winnipeg, Manitoba, Canada. The results confirmed that the proposed methodology was able to provide a higher overall accuracy and kappa coefficient than traditional stacking method, and also than all the individual RFs using each group. These accuracy metrics were also better than those of the RFs using the stacked features. Moreover, only the proposed methodology could achieve standard accuracy (F-score ≥85%) for all crop types in the study area. The visual comparison also demonstrated that the crop maps produced by the proposed methodology had more homogeneous, uniform appearances. Moreover, the mixed pixels of crop types, which abundantly existed in the traditional stacking and individual RFs̕ maps, were significantly eliminated.

Research paper thumbnail of Separability analysis of multifrequency SAR polarimetric features for land cover classification

Remote Sensing Letters, 2017

ABSTRACT Polarimetric Synthetic Aperture Radar (PolSAR) imagery can provide valuable observables ... more ABSTRACT Polarimetric Synthetic Aperture Radar (PolSAR) imagery can provide valuable observables at different frequencies for classification tasks. In this paper, we assessed separability rate of various polarimetric features in three frequencies of X-, C-, and L- bands. To this end, Jeffries–Matusita distance was firstly used to measure separability of each polarimetric feature in each frequency band. Random Forest classifier was then applied to map various land cover classes in study area. The classification outputs indicated that C-band results were better and more reliable than L-band results and L-band results were subsequently better than X-band results. These results were perfectly compatible with the results obtained by the separability analysis of multifrequency PolSAR features.

Research paper thumbnail of Enhanced decision tree ensembles for land-cover mapping from fully polarimetric SAR data

International Journal of Remote Sensing, 2017

ABSTRACT Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great pot... more ABSTRACT Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great potential for mapping and monitoring agro-environmental systems. Numerous polarimetric features can be extracted from these complex observations which may lead to improve accuracy of land-cover classification and object characterization. This article employed two well-known decision tree ensembles, i.e. bagged tree (BT) and random forest (RF), for land-cover mapping from PolSAR imagery. Moreover, two fast modified decision tree ensembles were proposed in this article, namely balanced filter-based forest (BFF) and cost-sensitive filter-based forest (CFF). These algorithms, designed based on the idea of RF, use a fast filter feature selection algorithms and two extended majority voting. They are also able to embed some solutions of imbalanced data problem into their structures. Three different PolSAR datasets, with imbalanced data, were used for evaluating efficiency of the proposed algorithms. The results indicated that all the tree ensembles have higher efficiency and reliability than the individual DT. Moreover, both proposed tree ensembles obtained higher mean overall accuracy (0.5–14% higher), producer’s accuracy (0.5–10% higher), and user’s accuracy (0.5–9% higher) than the classical tree ensembles, i.e. BT and RF. They were also much faster (e.g. 2–10 times) and more stable than their competitors for classification of these three datasets. In addition, unlike BT and RF, which obtained higher accuracy in large ensembles (i.e. the high number of DT), BFF and CFF can also be more efficient and reliable in smaller ensembles. Furthermore, the extended majority voting techniques could outperform the classical majority voting for decision fusion.

Research paper thumbnail of Performance Evaluation of Object-based and Pixel-based Building Detection Algorithms from Very High Spatial Resolution Imagery

Photogrammetric Engineering & Remote Sensing, 2014

This paper reviews and evaluates four building extraction algorithms including two pixel-based an... more This paper reviews and evaluates four building extraction algorithms including two pixel-based and two object-based methods using a diverse set of very high spatial resolution imagery. The applied images are chosen from different places (the cities of Isfahan, Tehran, and Ankara) and different sensors (QuickBird and GeoEye-1), which are diverse in terms of building shape, size, color, height, alignment, brightness, and density. The results indicate that the performance and the reliability of two object-based algorithms are better than pixel-based algorithms; about 10 percent to 15 percent better for the building detection rate and 6 percent to 10 percent better for the reliability rate. However, in some cases, the detection rate of pixel-based algorithms has been greater than 80 percent, which is a satisfactory result. On the other hand, segmentation errors can cause limitations and errors in the object-based algorithms, so that the commission error of object-based algorithms has been higher than pixel-based algorithms in some cases.

Research paper thumbnail of Enhancing Crop Classification Accuracy through Synthetic SAR-Optical Data Generation Using Deep Learning

ISPRS International Journal of Geo-Information

Crop classification using remote sensing data has emerged as a prominent research area in recent ... more Crop classification using remote sensing data has emerged as a prominent research area in recent decades. Studies have demonstrated that fusing synthetic aperture radar (SAR) and optical images can significantly enhance the accuracy of classification. However, a major challenge in this field is the limited availability of training data, which adversely affects the performance of classifiers. In agricultural regions, the dominant crops typically consist of one or two specific types, while other crops are scarce. Consequently, when collecting training samples to create a map of agricultural products, there is an abundance of samples from the dominant crops, forming the majority classes. Conversely, samples from other crops are scarce, representing the minority classes. Addressing this issue requires overcoming several challenges and weaknesses associated with the traditional data generation methods. These methods have been employed to tackle the imbalanced nature of training data. Nev...

Research paper thumbnail of Enhancing crop classification accuracy by synthetic SAR-Optical data generation using deep learning

ISPRS International Journal of Geo-Information, 2023

Crop classification using remote sensing data has emerged as a prominent research area in recent... more Crop classification using remote sensing data has emerged as a prominent research area in recent decades. Studies have demonstrated that fusing synthetic aperture radar (SAR) and optical images can significantly enhance the accuracy of classification. However, a major challenge in this field is the limited availability of training data, which adversely affects the performance of classifiers. In agricultural regions, the dominant crops typically consist of one or two specific types, while other crops are scarce. Consequently, when collecting training samples to create a map of agricultural products, there is an abundance of samples from the dominant crops, forming the majority classes. Conversely, samples from other crops are scarce, representing the minority classes. Addressing this issue requires overcoming several challenges and weaknesses associated with the traditional data generation methods. These methods have been employed to tackle the imbalanced nature of training data. Nevertheless, they still face limitations in effectively handling minority classes. Overall, the issue of inadequate training data, particularly for minority classes, remains a hurdle that the traditional methods struggle to overcome. In this research, we explore the effectiveness of a conditional tabular generative adversarial network (CTGAN) as a synthetic data generation method based on a deep learning network, for addressing the challenge of limited training data for minority classes in crop classification using the fusion of SAR-optical data. Our findings demonstrate that the proposed method generates synthetic data with a higher quality, which can significantly increase the number of samples for minority classes, leading to a better performance of crop classifiers. For instance, according to the G-mean metric, we observed notable improvements in the performance of the XGBoost classifier of up to 5% for minority classes. Furthermore, the statistical characteristics of the synthetic data were similar to real data, demonstrating the fidelity of the generated samples. Thus, CTGAN can be employed as a solution for addressing the scarcity of training data for minority classes in crop classification using SAR–optical data.

Research paper thumbnail of A random forest-based framework for crop mapping using temporal, spectral, textural and polarimetric observations

International Journal of Remote Sensing, 2019

Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great potential fo... more Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great potential for mapping and monitoring agro-environmental systems. Numerous polarimetric features can be extracted from these complex observations which may lead to improve accuracy of land-cover classification and object characterization. This article employed two well-known decision tree ensembles, i.e. bagged tree (BT) and random forest (RF), for land-cover mapping from PolSAR imagery. Moreover, two fast modified decision tree ensembles were proposed in this article, namely balanced filter-based forest (BFF) and cost-sensitive filter-based forest (CFF). These algorithms, designed based on the idea of RF, use a fast filter feature selection algorithms and two extended majority voting. They are also able to embed some solutions of imbalanced data problem into their structures. Three different PolSAR datasets, with imbalanced data, were used for evaluating efficiency of the proposed algorithms. The results indicated that all the tree ensembles have higher efficiency and reliability than the individual DT. Moreover, both proposed tree ensembles obtained higher mean overall accuracy (0.5–14% higher), producer’s accuracy (0.5–10% higher), and user’s accuracy (0.5–9% higher) than the classical tree ensembles, i.e. BT and RF. They were also much faster (e.g. 2–10 times) and more stable than their competitors for classification of these three datasets. In addition, unlike BT and RF, which obtained higher accuracy in large ensembles (i.e. the high number of DT), BFF and CFF can also be more efficient and reliable in smaller ensembles. Furthermore, the extended majority voting techniques could outperform the classical majority voting for decision fusion.

Research paper thumbnail of Presenting an extended evaluation framework for building detection algorithms using high spatial resolution images

This paper aims to provide an extended evaluation framework for building detection algorithms usi... more This paper aims to provide an extended evaluation framework for building detection algorithms using a diverse set of High Spatial Resolution (HSR) images. The HSR images utilized in this paper were chosen from different places and different sensors, and based on several important challenges in an urban area such as building alignment, density, shape, size, color, height, and imaging angle. The classical evaluation metrics such as detection rate, reliability, false positive rate, and overall accuracy only demonstrate the performance evaluation of an algorithm in relation to the buildings and cannot interpret the mentioned challenges. The extended evaluation framework proposed in this paper composed several extended metrics for performance evaluation of building detection algorithms in relation to these challenges in addition to the classical metrics. The paper intends to declare that the success or failure metrics of a building detection algorithm can have more varieties. In fact, a ...

Research paper thumbnail of Fully polarimetric synthetic aperture radar data classification using probabilistic and non-probabilistic kernel methods

European Journal of Remote Sensing, 2021

The data classification of fully polarimetric synthetic aperture radar (PolSAR) is one of the fav... more The data classification of fully polarimetric synthetic aperture radar (PolSAR) is one of the favourite topics in the remote sensing community. To date, a wide variety of algorithms have been utilized for PolSAR data classification, and among them kernel methods are the most attractive algorithms for this purpose. The most famous kernel method, i.e., the support vector machine (SVM) has been widely used for PolSAR data classification. However, until now, no studies to classify PolSAR data have been carried out using certain extended SVM versions, such as the least squares support vector machine (LSSVM), relevance vector machine (RVM) and import vector machine (IVM). Therefore, this work has employed and compared these four kernel methods for the classification of three PolSAR data sets. These methods were compared in two groups: the SVM and LSSVM as non-probabilistic kernel methods vs. the RVM and IVM as probabilistic kernel methods. In general, the results demonstrated that the SVM was marginally better, more accurate and more stable than the probabilistic kernels. Furthermore, the LSSVM performed much faster than the probabilistic kernel methods and its associated version, the SVM, with comparable accuracy.

Research paper thumbnail of The Use of Multiple Classifier Systems For Improving the Classification Accuracy of High–Dimensional Fully Polarimetric SAR Images

Journal of Geospatial Information Technology, 2014

A fully polarimetric synthetic aperture radar (POLSAR) image can provide a high-dimensional data.... more A fully polarimetric synthetic aperture radar (POLSAR) image can provide a high-dimensional data. This large amount of information can increase the overall accuracy of land-cover classification. But increasing the data dimensions if inadequately number of training samples may increase the complexity and cause the curse of dimensionality phenomenon. One of the strategies for solving this problem is the use of multiple classifier systems (MCS) that has the capability of divide and conquer to the large data as compared to the individual classifiers. In addition, some of MCS methods using the weak and unstable classifiers such as decision tree (DT) and neural network (NN) can obtain the high accuracy in high-dimensional data. The objective of this paper is also to use several popular MCS methods such as adaboost, bagging and random forests in order to improve the accuracy of land-cover classification from high-dimensional PolSAR images. The data used in this paper are Radarsat-2 image from San Francisco Bay and AIRSAR image of Flevoland. For classifying two these images, 69 polarimetric features were extracted from them. Two classifiers of DT and NN were chosen as the base classifiers of adaboost and bagging methods. In the next, the MCS methods were compared with the individual classifiers of DT and NN. The results indicated the higher overall accuracy of MCS methods between 5%-8% for classifying first image and 9%-16% for classifying second image. Even, the producer's accuracy and user's accuracy of MCS methods at all classes were more than the those of individual classifiers. So that at some classes, the difference was between 20% to even near 50%. These results confirmed that the MCS methods not only can produce higher overall accuracy at land-cover classification, but also they have the higher efficiency and reliability at discriminate individual classes.

Research paper thumbnail of Presenting A Feature Selection Method Based On Genetic Algorithm and Decision Tree For Classifying Fully Polarimetric SAR Images

Journal of Geospatial Information Technology, 2015

A fully polarimetric synthetic aperture radar (POLSAR) image can provide important polarimetric f... more A fully polarimetric synthetic aperture radar (POLSAR) image can provide important polarimetric features for land cover classification. These features can be the parameters obtained from scatering, covariance and coherency matrices, parameters extracted from target decomposition methods or both of them. In this paper, many polarimetric features are extracted from a POLSAR image. Then, with the use of Genetic Algorithm (GA) and Decision Tree (DT), a feature selection method based on the classification is presented. Afterwards, a comparative analysis is accomplished between DT classification with features selected from the proposed method and DT classification with all features. Moreover, the proposed method should be compared with the feature selection method of GA and Support Vector Machine (SVM). The results indicated that the accuracy of the proposed method (DT classification with the features selected from GA-DT algorithm) is nearly 3% higher than the ones of the DT classification with all features and it is approximately equal with the ones of the DT classification with the features selected from GA-SVM algorithm. However, the performance speed of the proposed method is approximately 5 times more than the ones of DT classification with the features selected from GA-SVM algorithm. As an another result, the features selected from the proposed method have a more success than the ones of two other methods at classifying the urban areas and vegetation classes.

Research paper thumbnail of MSMD: maximum separability and minimum dependency feature selection for cropland classification from optical and radar data

International Journal of Remote Sensing, 2018

ABSTRACT Cropland classification using optical and full polarimetric synthetic aperture radar (Po... more ABSTRACT Cropland classification using optical and full polarimetric synthetic aperture radar (PolSAR) images is a topic of considerable interest in the remote-sensing community. These two data sources can provide a diverse set of temporal, spectral, textural and polarimetric features which can be invaluable for cropland classification. However, some optical features or some radar features may have a relatively high correlation with other features. Hence, it seems to be necessary to choose the optimum features in order to reduce the dimensions of the data and to improve cropland classification accuracy. This article proposes a strategic feature selection method from a feature set of bitemporal RapidEye and Uninhabited Aerial Vehicle synthetic aperture radar (UAVSAR) images. The proposed method is designed to select the most relevant features and to remove redundant features based on the two concepts of separability and dependency. The proposed method is therefore referred to as maximum separability and minimum dependency (MSMD). For evaluating efficiency, MSMD and some well-known filter and wrapper feature selection methods are compared using a random forest classifier. Experimental tests confirmed that the classification results obtained from the MSMD feature selection method were more accurate than those achieved by filter methods. Moreover, they had an accuracy comparable to that of the results from the wrapper method. Furthermore, with regard to running time, MSMD operated as fast as the filter methods. It had a straightforward structure compared to the wrapper method, and as a result was faster than this method.

Research paper thumbnail of A random forest-based framework for crop mapping using temporal, spectral, textural and polarimetric observations

International Journal of Remote Sensing, 2019

Combining optical and polarimetric synthetic aperture radar (PolSAR) earth observations offers a ... more Combining optical and polarimetric synthetic aperture radar (PolSAR) earth observations offers a complementary data set with a significant number of spectral, textural, and polarimetric features for crop mapping and monitoring. Moreover, a temporal combination of both sources of information may lead to obtaining more reliable results compared to the use of single-time observations. In this paper, an operational framework based on the stacked generalization of random forest (RF), which efficiently employed bi-temporal observations of optical and radar data, was proposed for crop mapping. In the first step, various spectral, vegetation index, textural, and polarimetric features were extracted from both data sources and placed into several groups. Each group was classified separately using a single RF classifier. Then, several additional classification tasks were accomplished by another RF classifier. The earth observations used in this paper were collected by RapidEye satellites and the Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) system over an agricultural region near Winnipeg, Manitoba, Canada. The results confirmed that the proposed methodology was able to provide a higher overall accuracy and kappa coefficient than traditional stacking method, and also than all the individual RFs using each group. These accuracy metrics were also better than those of the RFs using the stacked features. Moreover, only the proposed methodology could achieve standard accuracy (F-score ≥85%) for all crop types in the study area. The visual comparison also demonstrated that the crop maps produced by the proposed methodology had more homogeneous, uniform appearances. Moreover, the mixed pixels of crop types, which abundantly existed in the traditional stacking and individual RFs̕ maps, were significantly eliminated.

Research paper thumbnail of Separability analysis of multifrequency SAR polarimetric features for land cover classification

Remote Sensing Letters, 2017

ABSTRACT Polarimetric Synthetic Aperture Radar (PolSAR) imagery can provide valuable observables ... more ABSTRACT Polarimetric Synthetic Aperture Radar (PolSAR) imagery can provide valuable observables at different frequencies for classification tasks. In this paper, we assessed separability rate of various polarimetric features in three frequencies of X-, C-, and L- bands. To this end, Jeffries–Matusita distance was firstly used to measure separability of each polarimetric feature in each frequency band. Random Forest classifier was then applied to map various land cover classes in study area. The classification outputs indicated that C-band results were better and more reliable than L-band results and L-band results were subsequently better than X-band results. These results were perfectly compatible with the results obtained by the separability analysis of multifrequency PolSAR features.

Research paper thumbnail of Enhanced decision tree ensembles for land-cover mapping from fully polarimetric SAR data

International Journal of Remote Sensing, 2017

ABSTRACT Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great pot... more ABSTRACT Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great potential for mapping and monitoring agro-environmental systems. Numerous polarimetric features can be extracted from these complex observations which may lead to improve accuracy of land-cover classification and object characterization. This article employed two well-known decision tree ensembles, i.e. bagged tree (BT) and random forest (RF), for land-cover mapping from PolSAR imagery. Moreover, two fast modified decision tree ensembles were proposed in this article, namely balanced filter-based forest (BFF) and cost-sensitive filter-based forest (CFF). These algorithms, designed based on the idea of RF, use a fast filter feature selection algorithms and two extended majority voting. They are also able to embed some solutions of imbalanced data problem into their structures. Three different PolSAR datasets, with imbalanced data, were used for evaluating efficiency of the proposed algorithms. The results indicated that all the tree ensembles have higher efficiency and reliability than the individual DT. Moreover, both proposed tree ensembles obtained higher mean overall accuracy (0.5–14% higher), producer’s accuracy (0.5–10% higher), and user’s accuracy (0.5–9% higher) than the classical tree ensembles, i.e. BT and RF. They were also much faster (e.g. 2–10 times) and more stable than their competitors for classification of these three datasets. In addition, unlike BT and RF, which obtained higher accuracy in large ensembles (i.e. the high number of DT), BFF and CFF can also be more efficient and reliable in smaller ensembles. Furthermore, the extended majority voting techniques could outperform the classical majority voting for decision fusion.

Research paper thumbnail of Performance Evaluation of Object-based and Pixel-based Building Detection Algorithms from Very High Spatial Resolution Imagery

Photogrammetric Engineering & Remote Sensing, 2014

This paper reviews and evaluates four building extraction algorithms including two pixel-based an... more This paper reviews and evaluates four building extraction algorithms including two pixel-based and two object-based methods using a diverse set of very high spatial resolution imagery. The applied images are chosen from different places (the cities of Isfahan, Tehran, and Ankara) and different sensors (QuickBird and GeoEye-1), which are diverse in terms of building shape, size, color, height, alignment, brightness, and density. The results indicate that the performance and the reliability of two object-based algorithms are better than pixel-based algorithms; about 10 percent to 15 percent better for the building detection rate and 6 percent to 10 percent better for the reliability rate. However, in some cases, the detection rate of pixel-based algorithms has been greater than 80 percent, which is a satisfactory result. On the other hand, segmentation errors can cause limitations and errors in the object-based algorithms, so that the commission error of object-based algorithms has been higher than pixel-based algorithms in some cases.

Research paper thumbnail of Enhancing Crop Classification Accuracy through Synthetic SAR-Optical Data Generation Using Deep Learning

ISPRS International Journal of Geo-Information

Crop classification using remote sensing data has emerged as a prominent research area in recent ... more Crop classification using remote sensing data has emerged as a prominent research area in recent decades. Studies have demonstrated that fusing synthetic aperture radar (SAR) and optical images can significantly enhance the accuracy of classification. However, a major challenge in this field is the limited availability of training data, which adversely affects the performance of classifiers. In agricultural regions, the dominant crops typically consist of one or two specific types, while other crops are scarce. Consequently, when collecting training samples to create a map of agricultural products, there is an abundance of samples from the dominant crops, forming the majority classes. Conversely, samples from other crops are scarce, representing the minority classes. Addressing this issue requires overcoming several challenges and weaknesses associated with the traditional data generation methods. These methods have been employed to tackle the imbalanced nature of training data. Nev...

Research paper thumbnail of Enhancing crop classification accuracy by synthetic SAR-Optical data generation using deep learning

ISPRS International Journal of Geo-Information, 2023

Crop classification using remote sensing data has emerged as a prominent research area in recent... more Crop classification using remote sensing data has emerged as a prominent research area in recent decades. Studies have demonstrated that fusing synthetic aperture radar (SAR) and optical images can significantly enhance the accuracy of classification. However, a major challenge in this field is the limited availability of training data, which adversely affects the performance of classifiers. In agricultural regions, the dominant crops typically consist of one or two specific types, while other crops are scarce. Consequently, when collecting training samples to create a map of agricultural products, there is an abundance of samples from the dominant crops, forming the majority classes. Conversely, samples from other crops are scarce, representing the minority classes. Addressing this issue requires overcoming several challenges and weaknesses associated with the traditional data generation methods. These methods have been employed to tackle the imbalanced nature of training data. Nevertheless, they still face limitations in effectively handling minority classes. Overall, the issue of inadequate training data, particularly for minority classes, remains a hurdle that the traditional methods struggle to overcome. In this research, we explore the effectiveness of a conditional tabular generative adversarial network (CTGAN) as a synthetic data generation method based on a deep learning network, for addressing the challenge of limited training data for minority classes in crop classification using the fusion of SAR-optical data. Our findings demonstrate that the proposed method generates synthetic data with a higher quality, which can significantly increase the number of samples for minority classes, leading to a better performance of crop classifiers. For instance, according to the G-mean metric, we observed notable improvements in the performance of the XGBoost classifier of up to 5% for minority classes. Furthermore, the statistical characteristics of the synthetic data were similar to real data, demonstrating the fidelity of the generated samples. Thus, CTGAN can be employed as a solution for addressing the scarcity of training data for minority classes in crop classification using SAR–optical data.

Research paper thumbnail of A random forest-based framework for crop mapping using temporal, spectral, textural and polarimetric observations

International Journal of Remote Sensing, 2019

Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great potential fo... more Fully polarimetric synthetic aperture radar (PolSAR) Earth Observations showed great potential for mapping and monitoring agro-environmental systems. Numerous polarimetric features can be extracted from these complex observations which may lead to improve accuracy of land-cover classification and object characterization. This article employed two well-known decision tree ensembles, i.e. bagged tree (BT) and random forest (RF), for land-cover mapping from PolSAR imagery. Moreover, two fast modified decision tree ensembles were proposed in this article, namely balanced filter-based forest (BFF) and cost-sensitive filter-based forest (CFF). These algorithms, designed based on the idea of RF, use a fast filter feature selection algorithms and two extended majority voting. They are also able to embed some solutions of imbalanced data problem into their structures. Three different PolSAR datasets, with imbalanced data, were used for evaluating efficiency of the proposed algorithms. The results indicated that all the tree ensembles have higher efficiency and reliability than the individual DT. Moreover, both proposed tree ensembles obtained higher mean overall accuracy (0.5–14% higher), producer’s accuracy (0.5–10% higher), and user’s accuracy (0.5–9% higher) than the classical tree ensembles, i.e. BT and RF. They were also much faster (e.g. 2–10 times) and more stable than their competitors for classification of these three datasets. In addition, unlike BT and RF, which obtained higher accuracy in large ensembles (i.e. the high number of DT), BFF and CFF can also be more efficient and reliable in smaller ensembles. Furthermore, the extended majority voting techniques could outperform the classical majority voting for decision fusion.