Journal of the Indian Society of Remote Sensing最新文献

In remote sensing field, the image retrieval is considered a complex task and attained higher attention, because of the data acquired from the earth observation satellites. An understanding of remote sensing images is obstructed because of the large amount of remote sensing images, lack of labeled samples, and complex contents. Content-based image retrieval made the powerful tool to mine huge remote sensing image databases. In content-based image retrieval, the query image is given for acquiring the images with identical visual content from the huge amount of remote sensing image database. In this research, the K-means pelican optimization algorithm is proposed for ensuring the search space reduction to enhance the retrieval of remote sensing images. The different feature extraction approaches such as Resnet-18, gray level co-occurrence matrix, Color moments, and local binary pattern are used to perform an effective feature extraction. Further, the feature transformation and neighborhood component analysis based feature selection is performed to transform the features into the similar significance and to select optimum features. Three different datasets such as Aerial Image Dataset, Remote Sensing-Image Classification Benchmark-256 and Wuhan University-Remote Sensing datasets are used to evaluate the proposed K-means pelican optimization algorithm. The proposed method is analyzed using precision, recall, F1-score and Average Normalized Modified Retrieval Rank. The existing research such as gabor-channel attention-ResNet, squeeze and excitation networks with ResNet50 and fused convolutional neural network-relevance feedback model are used to compare the K-means pelican optimization algorithm. The precision of the K-means pelican optimization algorithm for the Aerial Image Dataset dataset is 96.29% which is high when compared to the gabor-channel attention-ResNet, squeeze and excitation networks-ResNet50 and fused convolutional neural network- relevance feedback model.

The central part of Jharkhand, India, harbours a complex history shaped by ancient civilisations, notably Buddhism, Jainism, and Brahmanism, necessitating a meticulous identification of potential archaeological sites. This study employs a cutting-edge machine learning approach to predict the suitability of archaeological sites in the region, marking a significant evolution in the documentation of such sites. Machine learning-based integration of 12 geoenvironmental datasets using a random forest model reveals a nuanced spatial distribution of potential archaeological sites, categorised into four suitability zones: high, moderately high, moderately low, and low. The region with the best-anticipated suitability comprises around 20.33% of the research area, whereas the area with the lowest expected suitability comprises nearly 41.81%. High suitability zones, characterised by gentle terrain, open vegetation, fertile soils, and water proximity, suggest conditions conducive to human habitation and archaeological preservation. Conversely, low suitability zones exhibit rugged terrain, dense vegetation, poor soil quality, limited water availability, and remoteness from natural resources, indicating potential hindrances to human occupation and archaeological preservation. The model exhibited high predictive accuracy, as evidenced by the ROC–AUC score of 88.3%, enhancing its reliability. Specific locations within the study area demonstrate varying degrees of suitability, providing valuable insights for archaeological site management, cultural heritage preservation, and land-use planning, which will support the restoration and conservation plan of the heritage sites. Furthermore, this machine learning-based archaeological site prediction study underscores its potential applicability in historically rich regions globally, showcasing its significance in uncovering and preserving our shared human history.

Various Mars missions and telescopic remote observations have provided a comprehensive understanding of the complex atmospheric phenomenon and the structure of the Martian atmosphere. Several studies reported remote observation of layered clouds up to 100 km on Mars. Based on telescope data, observation of an unusual plume at an altitude of 200–250 km during March–April 2012 was reported. No such observation from any Mars orbiting spacecraft has been reported so far. Using data from the Mars Colour Camera onboard the Mars Orbiter Mission acquired on 4 January 2016, we report the occurrence of a high-altitude bright plume at the Martian evening terminator at an altitude of 260–300 km above the surface. The plume is observed at Meridiani Planum with a central location near 4.9°E and 9.5°S. Five images acquired within 10 min shows rapid variability in plume shape due to the movement of spacecraft and Mars. Preliminary analysis of MAVEN in-situ measurements shows an extremely disturbed solar wind plasma state during the plume observation time. We cautiously conclude that the formation of this high-altitude plume may result from interplanetary coronal mass ejection that occurred on 28 December 2015 that impacted Mars at around 3–4 January 2016 as confirmed by the analysis of simulation results and of in-situ solar wind data.

Global bathymetric models derived from satellite altimetry are important for studying the Earth’s oceans. However, the accuracy of these models can vary across different geographic regions. This study evaluates four widely used global bathymetric models ETOPO 2022, GEBCO 2023, SRTM15 + V2.5.5, and DTU18BAT in the Red Sea using 268,071 reference shipborne bathymetric measurements. The analysis compares the models’ depth estimates to the shipborne measurements across different depth ranges between 0 and 3000 m. The results show that overall, the GEBCO 2023 model provides the highest accuracy with the lowest standard deviation of 43.774 m and root mean square error of 43.929 m relative to shipborne data. The ETOPO 2022 model ranks second in accuracy with a standard deviation of 45.316 m and root mean square error of 45.345 m. The frequency distribution of residuals indicates that GEBCO 2023 and ETOPO 2022 models have the most precise depth predictions concentrated tightly around zero difference, while SRTM15 + V2.5.5 and DTU18BAT ones show broader spreads. There is no systematic depth over or under-predictions. Finally, the GEBCO 2023 and ETOPO 2022 models show good accuracy in the Red Sea, outperforming SRTM15 + V2.5.5 and DTU18BAT.

The unmanned aerial vehicles are used with LIDAR technology and the CNN method to detect damages to roads, buildings, and bridges. The Light detection and ranging (LIDAR) is used for mapping and capturing the damage to roads and buildings, and it is a 3D mapping. The convolutional neural network (CNN) method and deep learning method are used to properly research the damaged areas and depend on low- to high-level pattern detection. It is used in visual detection and shows consistently superior accuracy for spectrogram classifications. It collects the data from damaged areas and gives the information to the device. Here, the instructions are designed in Python. We use multisensory to detect the cracks and pits, and the damaged places will be detected using sensors and sent as a pronouncement. The images are captured by the LIDAR and processed according to the instructions given by the build programming language. It is used to reduce work time and make it highly efficient. It can detect the damages automatically on high buildings, bridges, and roads. It is mostly used in civil departments. The experimental results shows that the proposed model attained the maximum accuracy of 95.88%.

Wetlands provide a variety of habitats for different life forms and are essential to human survival. Sasthamkotta Lake, designated as a freshwater wetland ecosystem and a Ramsar Site of international importance is currently facing challenges of nutrient enrichment from the nearby land use features. This study utilizes the InVEST software’s Nutrient Delivery Ratio module coupled with Remote Sensing and GIS (Geographic Information System) to analyse the spatial distribution and temporal variation in the impact of land use on nutrient delivery in the watershed area of Sasthamkotta lake comprising of three panchayats namely Sasthamkotta, Mynagappally and West Kallada. The result reveals that settlement with vegetation followed by open land with vegetation and dense vegetation were the dominant land use classes as well as the key contributors of Total Phosphorus (TP) and Total Nitrogen (TN) in the watershed area. The value of TP exported (varies from 0 to 0.700 million tonnes/km) and that of TN (varies from 0 to 0.450 million tonnes/km) demonstrates that TP export was higher. This could be due to runoff from agricultural land and rubber plantations, discharge from nearby residences, water treatment plant and anthropogenic activities, particularly in the 100 m buffer zone of the periphery of the lake. It also highlights the internal water flow pattern within the lake, which indicates a groundwater recharge zone near the bund region underlining the significance of sustainable land-use planning and management strategies in the watershed.

An extensive investigation is conducted using remote sensing and gravity datasets to comprehend the volcanic eruptions and tectonic activity within the Aitken crater, farside of the Moon. M³ analyses indicate that the mare region dominates the clinopyroxene and represents the basaltic nature. The southern part of the crater floor exhibits enhanced FeO (11–15 wt%) and TiO₂ (2–5 wt%) percentages, indicating mare basalt material in conjunction with the spectral data. We observed intense mass-wasting features, various small-scale tectonic and volcanic structures on the crater walls and floor. We found lobate scarps near the mare basalts; however, the thickness of the mare basalts is low; therefore, there was not much subsidence and contraction produced by the mare basalts. As a result, the lobate scarps in the mare basalts of Aitken were probably caused by the Moon's thermal contraction. The GRAIL gravity anomalies indicate the existence of deep-seated subsurface material (i.e., magmatism that caused the mare to form on the crater floor) and a thick crust (30–40 km). Based on these integrated (compositional, morphological, and gravity) observations, we conclude that the floor of the crater is probably volcanic in origin, and the walls of the crater formed due to the impact melt crystallization. The wrinkle ridges that cut across minor impact craters and volcanic domes, horseshoe-shaped depressions, lobate scarps, and well-preserved dome structures indicate crater modification in later stages due to volcanic and tectonic activity. The eruptive activity in Aitken most likely began with an explosive cone-building stage, continued with lava eruptions from cones and fissures, and ended with a drain limited to the relatively deep lava ponded in the vents. Future research and analysis of the Aitken crater is particularly attractive because of its combination of impact and volcanic features.

Timely and accurate information on crop-sown areas during the kharif season (monsoon season in India) is crucial for early identification of drought-prone areas, enabling prompt intervention and mitigation measures to minimize adverse effects on crops and farmers. In this study, two approaches were attempted to estimate the in-season kharif sown area by the end of August using EOS-04 (RISAT-1A) data. Approach 1 utilizes the Coefficient of Variation (CV) of temporal Synthetic Aperture Radar (SAR) backscatter, while Approach 2 integrates optical data with the CV of SAR backscatter. The algorithm based on the temporal CV suggested that the variability of backscatter values over time, captured through temporal analysis, can be a key factor in identifying and delineating cropland areas. The CV of temporal HV backscatter data serves as an indicator of changes in vegetation cover or crop growth stages. In this study, CV values for settlement, forest, and fallow areas were observed to be 0.18, 0.17, and 0.19, respectively, while crops exhibited higher CV values of more than 0.4, which can be attributed to active crop growth. CV threshold optimization was carried out using Youden’s J Score statistical method. The optimal CV threshold value was observed to be 0.3, computed based on the temporal HV backscatter data from four study districts, which was further validated over two other districts. Accuracies of around 80% were achieved in both test and validation districts using the SAR only approach. Integration of optical data with SAR data led to improved overall accuracies, ranging from 85 to 89% in all test and validation districts. The findings suggest that CV analysis of backscatter values, complemented with optical data, can be a valuable tool for early discrimination between different land cover features, with croplands standing out due to their higher CV values attributed to the dynamic nature of crop growth. Using Youden’s J Score for threshold optimization adds statistical rigor to the methodology and demonstrates its potential for accurate in-season kharif sown area estimation for its operational use over large areas.

Image fusion methods may lose their ability to retain crucial image information when faced with suboptimal conditions, such as poor contrast, excessive noise, or intense illumination, leading to the loss of valuable image features. In this work, an improved CsdFusion algorithm is proposed to increase the visibility of infrared targets in fused images. Firstly, to accomplish clear background textures and structural information, a hybrid image decomposition model combining LatLRR and NSST is established. This process entails the division of the original infrared and visible images into low-rank components (base layers) and salient components (saliency layers) through the Latent Low-Rank Representation (LatLRR) approach. Subsequently, the base layers of both the infrared and visible images undergo the Non-Subsampled Shearlet Transform (NSST), decomposing them into high-frequency and low-frequency layers. The processed high-frequency and low-frequency layers are then subjected to inverse NSST to obtain the fused base layer, ensuring that the fused image retains maximum background information while effectively filtering noise. Secondly, to identify and extract the most significant regions or features in infrared images, the Central-contrast priori Saliency Map (CSM) algorithm is applied. This algorithm calculates the central prior saliency value using Harris corners and the contrast prior saliency value using guided filtering and background suppression. It then combines these two prior saliency values using a feature compensation strategy to compute the infrared saliency map. To validate the effectiveness of the proposed algorithm, comparative evaluation studies on benchmark open datasets are carried out. The results thus obtained through the proposed algorithm demonstrate superior performance in both subjective and objective experiments, generating fused images that not only preserve the crucial details and characteristics of both infrared and visible images but also reflect significant enhancement in visibility and discriminability of target objects, outperforming 10 state-of-the-art image fusion algorithms.

The global decline in air quality, attributed to pollutants from various sources such as fossil fuel usage, industrial expansion, and heightened commercial activities, underscores the importance of monitoring and forecasting air quality levels. This study delves into 3 years of daily particulate matter data spanning the pre-COVID (2019), COVID-era (2020), and post-COVID (2021) periods across thirty-seven monitoring stations in Delhi. Prior to analysis, the dataset underwent preprocessing to address missing and outlier values. Analysis of the dataset aimed to discern pollutant trends across stations and timeframes, identifying influential factors such as air temperature, surface pressure, and precipitation for modeling particulate matter concentrations. An Artificial Neural Network employing backpropagation was utilized for modeling. Training the model with 80% of the dataset, the remaining 20% served as the test dataset. Validation of the model's performance utilized standard statistical metrics including R², r, root mean square error, and mean absolute error. Notably, the R² for the training dataset were 0.82 and 0.84 and r for training dataset were 0.90 & 0.91 for PM 10 and PM 2.5, respectively. While the R² for the test dataset were 0.78 and 0.79, r values for the test dataset stood at 0.88 for both PM 10 and PM 2.5. Furthermore, the model facilitated upscaling of observations to a spatial scale, broadening the scope of observations via simulations to enhance regional understanding.