Atmospheric Pollution Research (13091042)15(7)
Aerosol Optical Depth (AOD) across various altitudes is crucial for gaining a comprehensive understanding of aerosol dynamics. However, current methodologies utilizing passive remote sensing and active sensors have limitations in providing precise vertical coverage. In our methodology, we introduce a Seasonal-Independent model employing Machine Learning (ML) algorithms to retrieve AOD values at both 1.5 km and 3 km layers. The propose approach is assessed the performance of various ML algorithms, including XGBoost, Random Forest (RF), Multilayer Perceptron (MLP), and Support Vector Regression (SVR). Remarkably, our study successfully overcame seasonal constraints, yielding impressive R2 values of 0.94, 0.93, 0.93, and 0.87 for the 1.5 km layer, and 0.83, 0.79, 0.82, and 0.78 for the 3 km layer across the mentioned models for 2017, 2018 and 2019 data. Evaluating the proposed Seasonal-Independent XGBoost model against CALIOP AOD values for the 2020 data, we observed substantial agreement with R2 values of 0.93 and 0.81, and minimal RMSE values of 0.002 and 0.004 for the AODs at 1.5 km and 3 km, respectively. Furthermore, a comparative analysis of trends between estimated and CALIOP AODs revealed a strong resemblance in both altitude layers. © 2024 Turkish National Committee for Air Pollution Research and Control
Earth Science Informatics (18650473)16(3)pp. 2529-2543
Classifying urban land use/cover types poses significant challenges due to the complex and heterogeneous nature of urban landscapes. Recent years have witnessed notable advancements in land use/cover classification, driven by improvements in classification methods and the utilization of data from multiple sources. Deep learning networks, especially, have gained prominence in various image analysis tasks, including land use/cover classification. However, when it comes to urban areas, the classification of urban land use/cover encounters additional obstacles, including the complexity of classes, limited training data, and the presence of numerous urban categories. To overcome the limitations arising from similar classes and insufficient training data, we propose a novel approach that integrates hyperspectral and LiDAR data through a Conditional Generative Adversarial Network (CGAN) for semantic segmentation. Our methodology leverages the UNet + + generator and PatchGAN discriminator to achieve accurate segmentation. The CGAN-generated segmented images are then processed by a fully connected neural network (FCN) to classify 20 land use/cover classes. By validating our approach on the 2018 GRSS Data Fusion dataset, our study demonstrates its exceptional operational performance. The CGAN architecture outperforms previous algorithms in terms of class diversity and training data volume. By generating synthetic data that closely resembles the ground truth, the CGAN enhances the classification performance. Clear visual distinctions are observed among various urban features, such as vegetation, trees, buildings, roads, and cars. Classes associated with healthy grass, stressed grass, bare earth, and stadium seats achieve high accuracy. However, road and railway classes exhibit poorer performance due to their similarity with sidewalk, highway, major thoroughfare, and crosswalk classes. Overall, our study showcases a significant improvement in classification accuracy, achieving an approximate accuracy of 96.98% compared to the winning articles presented in the 2018 competition, which achieved accuracies of 64.95% and 76.54%, respectively. This improvement in accuracy can be attributed to the effective extraction and combination of high and low-level urban land cover/land use features within our proposed architecture. © 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.
Earth Science Informatics (18650473)16(1)pp. 753-771
One of the techniques for estimating the surface particle concentration with a diameter of fewer than 2.5 micrometers (PM2.5) is using aerosol optical depth (AOD) products. Different AOD products are retrieved from various satellite sensors, like MODIS and VIIRS, by various algorithms, such as Deep Blue and Dark Target. Therefore, they don’t have the same accuracy and spatial resolution. Additionally, the weakness of algorithms in AOD retrieval reduces the spatial coverage of products, particularly in cloudy or snowy areas. Consequently, for the first time, the present study investigated the possibility of fusing AOD products from observations of MODIS and VIIRS sensors retrieved by Deep Blue and Dark Target algorithms to estimate PM2.5 more accurately. For this purpose, AOD products were fused by machine learning algorithms using different fusion strategies at two levels: the data level and the decision level. First, the performance of various machine learning algorithms for estimating PM2.5 using AOD data was evaluated. After that, the XGBoost algorithm was selected as the base model for the proposed fusion strategies. Then, AOD products were fused. The fusion results showed that the estimated PM2.5 accuracy at the data level in all three metrics, RMSE, MAE, and R2, was improved (R2= 0.64, MAE= 9.71μgm3, RMSE= 13.51μgm3). Despite the simplicity and lower computational cost of the data level fusion method, the spatial coverage did not improve considerably due to eliminating poor quality data through the fusion process. Afterward, the fusion of products at the decision level was followed in eleven scenarios. In this way, the best result was obtained by fusing Deep Blue products of MODIS and VIIRS sensors (R2= 0.81, MAE= 7.38μgm3, RMSE= 10.08μgm3). Moreover, in this scenario, the spatial coverage was improved from 77% to 84%. In addition, the results indicated the significance of the optimal selection of AOD products for fusion to obtain highly accurate PM2.5 estimations. © 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.
Atmospheric Environment (13522310)292
Remote Sensing observations have enormous advantages in aerosol studies since aerosols' space and time variation. MODIS and CALIOP are two independent instruments with different design principles that provide aerosol optical depth (AOD) retrievals and scan the same points on the Earth's surface within a 2-min interval. Due to predefined aerosol models and fixed vertical profiles in the MODIS algorithm and AOD CALIOP resolution, MODIS and CALIOP cannot give suitable spatial-temporal coverage in related studies. This paper proposes a method based on Bayesian networks to retrieve the AODs by the synergy of CALIOP and MODIS in two vertical layers, 1.5 and 3 km. We applied the Bayesian network for three days in 2018 over the Persian Gulf. The overall analyses reveal that estimated AOD by the seasonal networks correlates with obtained retrieval CALIOP AODs. The correlation values, 0.94 and 0.84, are obtained for the first layer in the summer and winter. These values for the second layer are 0.88 and 0.82. The observed differences in the estimated AOD with the actual measured AOD values and the overall correlation results demonstrate that the proposed networks are sufficient to provide accurate AODs in the two, 1.5, and 3 km vertical layers. According to the experimental results, the layering MAIAC AOD product becomes more suitable for monitoring and studying aerosol phenomena by applying the proposed networks. © 2022 Elsevier Ltd
Seydgar, M.,
Alizadeh naeini, A.,
Zhang, M.,
Li, W.,
Sattari, M. Remote Sensing (20724292)11(7)
Nowadays, 3-D convolutional neural networks (3-D CNN) have attracted lots of attention in the spectral-spatial classification of hyperspectral imageries (HSI). In this model, the feed-forward processing structure reduces the computational burden of 3-D structural processing. However, this model as a vector-based methodology cannot analyze the full content of the HSI information, and as a result, its features are not quite discriminative. On the other hand, convolutional long short-term memory (CLSTM) can recurrently analyze the 3-D structural data to extract more discriminative and abstract features. However, the computational burden of this model as a sequence-based methodology is extremely high. In the meanwhile, the robust spectral-spatial feature extraction with a reasonable computational burden is of great interest in HSI classification. For this purpose, a two-stage method based on the integration of CNN and CLSTM is proposed. In the first stage, 3-D CNN is applied to extract low-dimensional shallow spectral-spatial features from HSI, where information on the spatial features are less than that of the spectral information; consequently, in the second stage, the CLSTM, for the first time, is applied to recurrently analyze the spatial information while considering the spectral one. The experimental results obtained from three widely used HSI datasets indicate that the application of the recurrent analysis for spatial feature extractions makes the proposed model robust against different spatial sizes of the extracted patches. Moreover, applying the 3-D CNN prior to the CLSTM efficiently reduces the model's computational burden. The experimental results also indicated that the proposed model led to a 1% to 2% improvement compared to its counterpart models. © 2019 by the authors.
Sattari, M.,
Babadi, M.,
Iran Pour S.,
Babadi, M.,
Iran Pour S. CTIT workshop proceedings series (16821750)42(4/W18)pp. 147-152
Precise measurements of forest trees is very important in environmental protection. Measuring trees parameters by use of ground-based inventories is time and cost consuming. Employing advanced remote sensing techniques to obtain forest parameters has recently made a great progress step in this research area. Among the information resources of the study field, full waveform LiDAR data have attracted the attention of researchers in the recent years. However, decomposing LiDAR waveforms is one of the challenges in the data processing. In fact, the procedure of waveform decomposition causes some of the useful information in waveforms to be lost. In this study, we aim to investigate the potential use of non-decomposed full waveform LiDAR features and its fusion with optical images in classification of a sparsely forested area. We consider three classes including i) ground, ii) Quercus wislizeni and iii) Quercus douglusii for the classification procedure. In order to compare the results, five different strategies, namely i) RGB image, ii) common LiDAR features, iii) fusion of RGB image and common LiDAR features, iv) LiDAR waveform structural features and v) fusion of RGB image and LiDAR waveform structural features have been utilized for classifying the study area. The results of our study show that classification via using fusion of LiDAR waveform features and RGB image leads to the highest classification accuracy. © 2019 M. Babadi et al.
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (21511535)12(2)pp. 675-684
Although scale invariant feature transform (SIFT) based algorithms have a wide range of applications in remote sensing panchromatic image matching, but minor changes in the contrast threshold can bring about drastic changes in the image matching quality. In order to effectively improve the matching quality of the SIFT detector, this paper proposes an adaptive contrast threshold-SIFT (ACT-SIFT) procedure. The ACT-SIFT method set forth herein harmoniously calculates two sought after contrast thresholds for the target and the reference images separately an objective achieved through minimizing certain proposed criterion. As an introductory step, the entropy in the keypoints (EiK) is defined as a control or balancing parameter being estimated in terms of the entropy of the image and the number of the keypoints. To that end, the candidate keypoints are first extracted from the scale space of SIFT method by applying the initial contrast threshold for both images. Then, the contrast threshold values are modified iteratively for both images to reach the needed values by minimizing the relative difference between the EiK of the target and the reference images. Next, in agreement with the threshold values obtained, the keypoints are extracted with the help of the SIFT keypoint detector. The correct matching pairs are also created using the matches acquired through the SIFT descriptor. The results obtained by applying our proposed approach promise boosted matching pairs in remote sensing image correspondences, for it extracts the keypoints in a robust manner by simultaneously checking the information content of the reference and the target images. © 2018 IEEE.
International Journal of Remote Sensing (13665901)39(3)pp. 684-703
Multiresolution segmentation (MRS) is one of the most commonly used image segmentation algorithms in the remote-sensing community. This algorithm has three user-defined parameters: scale, shape, and compactness. The scale parameter (SP) is the most crucial one in determining the average size of the image segments generated. Since setting this parameter typically requires a trialand- error process, automatically estimating it can expedite the segmentation process. However, most of automatic approaches are still iterative and can lead to a time-consuming process. In this article, we propose a new, non-iterative framework for estimating the SP with an emphasis on extracting individual urban buildings. The basis of the proposed method is to investigate the feasibility of associating the size of urban buildings with a corresponding ‘optimal’ (or at least reasonable) SP using an explicit mathematical equation. Using the proposed method, these two variables are related to each other by constructing a mathematical (regression) model. In this framework, the independent variables were chosen to be the typical size of buildings in a given urban area and the spatial resolution of the image under consideration; and the dependent variable was chosen to be the corresponding optimal SP. To assess the potential of the proposed approach, two regression models that yielded explicit equations (i.e. degree-2 polynomial (DP), and regression tree (RT)) were employed. In addition, as a sophisticated and versatile nonlinear model, support vector regression (SVR) was utilized to further measure the performances of DP and RT models compared with it. According to the comparisons, the DP model was selected as a representative of the proposed approach. In the end, to evaluate the proposed methodology, we also compared the results derived from the DP model with those derived from the Estimation of Scale Parameter (ESP) tool. Based on our experiments, not only did the DP model produce acceptable results, but also it outperformed ESP tool in this study for extracting individual urban buildings. © 2017 Informa UK Limited, trading as Taylor and Francis Group. All rights reserved.
CTIT workshop proceedings series (16821750)42(4W4)pp. 111-116
This paper presents an automatic method to extract road centerline networks from high and very high resolution satellite images. The present paper addresses the automated extraction roads covered with multiple natural and artificial objects such as trees, vehicles and either shadows of buildings or trees. In order to have a precise road extraction, this method implements three stages including: classification of images based on maximum likelihood algorithm to categorize images into interested classes, modification process on classified images by connected component and morphological operators to extract pixels of desired objects by removing undesirable pixels of each class, and finally line extraction based on RANSAC algorithm. In order to evaluate performance of the proposed method, the generated results are compared with ground truth road map as a reference. The evaluation performance of the proposed method using representative test images show completeness values ranging between 77% and 93%.
CTIT workshop proceedings series (16821750)41pp. 337-343
Airborne LiDAR (Light Detection and Ranging) data have a high potential to provide 3D information from trees. Most proposed methods to extract individual trees detect points of tree top or bottom firstly and then using them as starting points in a segmentation algorithm. Hence, in these methods, the number and the locations of detected peak points heavily effect on the process of detecting individual trees. In this study, a new method is presented to extract individual tree segments using LiDAR points with 10cm point density. In this method, a two-step strategy is performed for the extraction of individual tree LiDAR points: finding deterministic segments of individual trees points and allocation of other LiDAR points based on these segments. This research is performed on two study areas in Zeebrugge, Bruges, Belgium (51.33° N, 3.20° E). The accuracy assessment of this method showed that it could correctly classified 74.51% of trees with 21.57% and 3.92% under-and over-segmentation errors respectively.
CTIT workshop proceedings series (16821750)2016pp. 941-946
Anyone knows that sudden catastrophes can instantly do great damage. Fast and accurate acquisition of catastrophe information is an essential task for minimize life and property damage. Compared with other ways of catastrophe data acquisition, UAV based platforms can optimize time, cost and accuracy of the data acquisition, as a result UAVs' data has become the first choice in such condition. In this paper, a novel and fast strategy is proposed for registering and mosaicking of UAVs' image data. Firstly, imprecise image positions are used to find adjoining frames. Then matching process is done by a novel matching method. With keeping Sift in mind, this fast matching method is introduced, which uses images exposure time geometry, SIFT point detector and rBRIEF descriptor vector in order to match points efficiency, and by efficiency we mean not only time efficiency but also elimination of mismatch points. This method uses each image sequence imprecise attitude in order to use Epipolar geometry to both restricting search space of matching and eliminating mismatch points. In consideration of reaching to images imprecise attitude and positions we calibrated the UAV's sensors. After matching process, RANSAC is used to eliminate mismatched tie points. In order to obtain final mosaic, image histograms are equalized and a weighted average method is used to image composition in overlapping areas. The total RMSE over all matching points is 1.72 m.
Photogrammetric Record (14779730)27(139)pp. 330-359
In this paper, a multi-resolution hybrid approach is proposed for the reconstruction of building models from point clouds of lidar data. The detection of the main roof planes is obtained through a polyhedral approach, whereas the models of appended parts, in this case the dormers, are reconstructed by adopting a model-driven approach. Clustering of the roof points in a multi-resolution space is based on the fuzzy c-mean in the polyhedral section of this hybrid approach. A weighted plane algorithm is developed in order to determine the planes of each cluster. The verification of planes between multi-resolution spaces adopts a method based on a least squares support vector machine that, in the model-driven section, is applied for detecting types of projecting structures. A method is then developed to determine the dormer models' parameters. Finally, the detection of boundary roof lines is obtained through a customised fuzzy Hough transform. The paper outlines the concept of the algorithms and the processing chain, and illustrates the results obtained by applying the technique to buildings of different complexities. © 2012 The Authors. The Photogrammetric Record © 2012 The Remote Sensing and Photogrammetry Society and Blackwell Publishing Ltd.
Sattari, M.,
Shahbazi, M.,
Homayouni, S.,
Saadatseresht, M.,
Shahbazi, M.,
Sattari, M.,
Homayouni, S.,
Saadatseresht, M. CTIT workshop proceedings series (16821750)39pp. 51-56
Recent advances in positioning techniques have made it possible to develop Mobile Mapping Systems (MMS) for detection and 3D localization of various objects from a moving platform. On the other hand, automatic traffic sign recognition from an equipped mobile platform has recently been a challenging issue for both intelligent transportation and municipal database collection. However, there are several inevitable problems coherent to all the recognition methods completely relying on passive chromatic or grayscale images. This paper presents the implementation and evaluation of an operational MMS. Being distinct from the others, the developed MMS comprises one range camera based on Photonic Mixer Device (PMD) technology and one standard 2D digital camera. The system benefits from certain algorithms to detect, recognize and localize the traffic signs by fusing the shape, color and object information from both range and intensity images. As the calibrating stage, a self-calibration method based on integrated bundle adjustment via joint setup with the digital camera is applied in this study for PMD camera calibration. As the result, an improvement of 83% in RMS of range error and 72% in RMS of coordinates residuals for PMD camera, over that achieved with basic calibration is realized in independent accuracy assessments. Furthermore, conventional photogrammetric techniques based on controlled network adjustment are utilized for platform calibration. Likewise, the well-known Extended Kalman Filtering (EKF) is applied to integrate the navigation sensors, namely GPS and INS. The overall acquisition system along with the proposed techniques leads to 90% true positive recognition and the average of 12 centimetres 3D positioning accuracy.
2025 29th International Computer Conference, Computer Society of Iran, CSICC 2025pp. 6005-6008
In this paper, a method is presented to recognition and modeling of three types of dormers from lidar data. The input data of this proposed algorithm involves raw roof lidar data, a regular grid of lidar data and an initial building model without superstructures. This proposed method is modular. The first stage provides a recognition type of dormers via a support vector machine. The second stage reconstructs the dormer models. Experiments show the efficiency of the proposed method. © 2012 IEEE.
Sattari, M.,
Shahbazi, M.,
Homayouni, S.,
Saadatseresht, M.,
Shahbazi, M.,
Sattari, M.,
Homayouni, S.,
Saadatseresht, M. 2025 29th International Computer Conference, Computer Society of Iran, CSICC 2025pp. 148-155
This paper describes a method for detecting and recognizing traffic signs by integrating the range and intensity images of a Time-of-flight camera, based on Photonic Mixer Device (PMD) technology, with images of a standard digital camera. The reflectivity of signs surfaces along with background suppression ability and active sensing of the PMD camera make the signs sharply visible in intensity images. Besides the image descriptors, utilizing the object-based information provides robust and reliable detection and recognition. The overall acquisition system and proposed technique overcome the conventional illumination, disorientation and scaling problems in detection and recognition process. The method of this paper is implemented and evaluated on data acquired by a multi-sensor mobile mapping system.
Shahbazi, M.,
Homayouni, S.,
Saadatseresht, M.,
Sattari, M. Sensors (14248220)11(9)pp. 8721-8740
Time-of-flight cameras, based on Photonic Mixer Device (PMD) technology, are capable of measuring distances to objects at high frame rates, however, the measured ranges and the intensity data contain systematic errors that need to be corrected. In this paper, a new integrated range camera self-calibration method via joint setup with a digital (RGB) camera is presented. This method can simultaneously estimate the systematic range error parameters as well as the interior and external orientation parameters of the camera. The calibration approach is based on photogrammetric bundle adjustment of observation equations originating from collinearity condition and a range errors model. Addition of a digital camera to the calibration process overcomes the limitations of small field of view and low pixel resolution of the range camera. The tests are performed on a dataset captured by a PMD[vision]-O3 camera from a multi-resolution test field of high contrast targets. An average improvement of 83% in RMS of range error and 72% in RMS of coordinate residual, over that achieved with basic calibration, was realized in an independent accuracy assessment. Our proposed calibration method also achieved 25% and 36% improvement on RMS of range error and coordinate residual, respectively, over that obtained by integrated calibration of the single PMD camera. © 2011 by the authors; licensee MDPI, Basel, Switzerland.
CTIT workshop proceedings series (16821750)37pp. 823-827
In this paper we present and develop a set of algorithms, mostly based on morphological operators, for automatic colonic polyp detection applied to computed tomography (CT) scans. Initially noisy images are enhanced using Morphological Image Cleaning (MIC) algorithm. Then the colon wall is segmented using region growing followed by a morphological grassfire operation. In order to detect polyp candidates we present a new Automatic Morphological Polyp Detection (AMPD) algorithm. Candidate features are classified as polyps and non-polyps performing a novel Template Matching Algorithm (TMA) which is based on Euclidean distance searching. The whole technique achieved 100% sensitivity for detection of polyps larger than 10 mm and 81.82% sensitivity for polyps between 5 to 10 mm and expressed relatively low sensitivity (66.67%) for polyps smaller than 5 mm. The experimental data indicates that our polyp detection technique shows 71.73% sensitivity which has about 10 percent improvement after adding the noise reduction algorithm.
Survey Review (17522706)38(296)pp. 165-173
It is possible to use single frequency GPS receivers to estimate the Total Electron Content (TEC). In this research, we improved an algorithm presented by Giffard [2], that is based on a least squares solution. We investigated the effect of the use of different weights (elevation of satellites, signal to noise ratio, combination of elevation and signal to noise ratio) and different block sizes on TEC estimates. We found that these parameters had a significant impact on TEC estimates based on this algorithm. Our research is based on observations at the GPS site of the Esfahan University made with single frequency 12-channel Leica System 500 receivers.
Proceedings of SPIE - The International Society for Optical Engineering (1996756X)3957pp. 398-402
Virtual Reality (VR) is a possible which brings users to the reality by computer and Virtual Environment (VE) is a simulated world which takes users to any points and directions of the object. VR and VE can be very useful if accurate and precise data are used, and allows users to work with realistic model. Photogrammetry is a technique which is able to collect and provide accurate and precise data for building 3D model in a computer. Data can be collected from various sensors and cameras, and methods of data collector are vary based on the method of image acquiring. Indeed VR includes real-time graphics, three-dimensional model, and display and it has application in the entertainment industry, flight simulators, industrial design. Above definitions describe the relationship between VR and VE with photogrammetry. This paper describes a reliable and precis method of data acquiring based on close range photogrammetry for building a VR model. The purpose of this project is to make a real possibility for seismic designers to investigate all effects of shaking on a real building. Minar Gonban is an ancient building with two amazing minarets at Esfehan IRAN. While one of them was shaken the second one started to shake. The project is fulfilled on this building because building simply can be shaken and its effects can be investigated. The building was photographed by multiple movie cameras and photo cameras. Sequence images were restored in a computer for creating sequence models of building. A VR model is builded based on extracted data from photogrammetry images. The developed VR model is precise and reliable and provides real possibility for users to investigate the effects of shaking on the building. The developed VR model is based on real data. The results verify a reliable VR can be useful for human life because one of its application can help to investigate effects of earthquake on the building and duce its casualty.
