Multispectral Bands Research Articles

Multifunctional-hierarchical flexibility metasurfaces for multispectral compatible camouflage of microwave, infrared and visible.

The prevalent use of multispectral detection technology makes single-band camouflage devices ineffective, and the investigation of technology for camouflage that combines multispectral bands becomes urgent. The multifunctional-hierarchical flexibility metasurfaces (MHFM) for multispectral compatible camouflage of microwave, infrared, and visible, is proposed, fabricated, and measured. MHFM is primarily composed of an infrared shielding layer (IRSL), a radar absorbing layer (RAL), and a visible color layer (VCL). Among them, IRSL can block thermal infrared detection, and RAL can efficiently absorb microwave band electromagnetic (EM) waves. The VLC can display black (below 28°C), purple (28°C∼31°C), green (31°C∼33°C), and yellow (above 33°C) at different temperatures to achieve visible camouflage. Simulation results show that MHFM can achieve absorption higher than 90% in the 2.9∼13.9 GHz microwave band. Theoretically, the emissivity of MHFM in the infrared spectral range 3∼14 µm is less than 0.34. In addition, the MHFM consists of high-temperature-resistant materials that can be used normally at temperatures up to 175°C, providing excellent high-temperature stability. The measurement results show that the camouflage performance of the MHFM is in excellent agreement with the proposed theory. This study proposes a new method for multispectral camouflage that has broad engineering applications.

Multispectral and Thermal Sensors Onboard UAVs for Heterogeneity in Merlot Vineyard Detection: Contribution to Zoning Maps

This work evaluated the ability of UAVs to detect field heterogeneity and their influences on vineyard development in Yepes (Spain). Under deficit irrigation, vine growth and yield variability are influenced by soil characteristics such as water holding capacity (WHC). Over two irrigation seasons (2021–2022), several vegetation indices (VIs) and parameters of vegetative growth and yield were evaluated in two field zones. Multispectral and thermal information was obtained from bare soils. The water availability showed annual differences; it was reduced by 49% in 2022 compared to 2021, suggesting that no significant differences were found for the parameters studied. The zone with higher WHC also had the higher vegetative growth and yield in 2021. This agreed with the significant differences among the VIs evaluated, especially the ratio vegetation index (RVI). Soil multispectral and thermal bands showed significant differences between zones in both years. This indicated that the soil spectral and thermal characteristics could provide more reliable information for zoning than vine vegetation itself, as they were less influenced by climatic conditions between years. Consequently, UAVs proved to be valuable for assessing spatial and temporal heterogeneity in the monitoring of vineyards. Soil spectral and thermal information will be essential for zoning applications due to its consistency across different years, enhancing vineyard management practices.

Separate and Combined Effective Coding of Bit Planes of Grayscale Images

Currently, an approach involving a coder with a combined structure for compressing images combining several different coders, the system for connecting them to various bit planes, and the control system for these connections have not been studied. Thus, there is a need to develop a structure and study the effectiveness of a combined codec for compressing images of various types without loss in the spatial domain based on arithmetic and (Run-Length Encoding) RLE-coding algorithms. The essence of separate effective coding is to use independent coders of the same type or one coder connected to the planes alternately in order to compress the higher and lower bit planes of the image or their combinations. In this paper, the results of studying the effectiveness of using a combination of arithmetic and RLE coding for several types of images are presented. As a result of developing this structure, the effectiveness of combined coding for compressing the differences in the channels of hyperspectral images (HSI) has been established, as hyperspectral images consist of multi-spectral bands, instead of just the typical three bands (RGB) or (YCbCr) found in regular images. Where, each pixel in a hyperspectral image represents the entire spectrum of light reflected by the object or scene at that particular location.

Improving Forest Detection Using Machine Learning and Remote Sensing: A Case Study in Southeastern Serbia

Vegetation plays an active role in ecosystem dynamics, and monitoring its patterns and changes is vital for effective environmental resource management. This study explores the possibility of machine learning techniques and remote sensing data to improve the accuracy of forest detection. The research focuses on the southeastern part of the Republic of Serbia as a case study area, using Sentinel-2 multispectral bands. The study employs publicly accessible satellite data and incorporates different vegetation indices to improve classification accuracy. The main objective is to examine the practicability of expanding the input parameters for forest detection using a machine learning approach. The classification process is performed by employing support vector machines (SVM) algorithm and utilising the SVM module in the scikit-learn package. The results demonstrate that including vegetation indices alongside the multispectral bands significantly improves the accuracy of vegetation detection. A comprehensive assessment reveals an overall classification accuracy of up to 99.01% when the selected vegetation indices (MCARI, RENDVI, NDI45, GNDVI, NDII) are combined with the Sentinel-2 bands. This research highlights the potential of machine learning and remote sensing in forest detection and monitoring. The findings underscore the importance of incorporating vegetation indices to enhance classification accuracy using the Python programming language. The study’s outcomes provide valuable insights for environmental resource management and decision-making processes, particularly in regions with diverse forest ecosystems.

High performance infrared selective emissivity film tailored for thermal-stable camouflage

A polarization-independent broadband infrared selective absorber/emitter (ISAE) based on multilaminar architecture is proposed and demonstrated. The salient features are that it has both low average emissivity of less than 0.1 in two atmospheric windows (3–5μm and 8–14μm) and otherwise high average emissivity of more than 0.8 in two non-atmospheric windows (2.5–3μm and 5–8μm), which renders it tailored infrared camouflage performance with thermal stability. Meanwhile, there is sharp narrowband absorption around at 10.6μm, which allows it to additionally possess laser camouflage performance. The comprehensive dependence of multispectral selective emissivity properties on the structural parameters, the polarization and incident angle of incoming excitation are analyzed and the underlying physical mechanisms are explored. It is found that the selective absorption/emissivity in band 5–8μm is originated from the fundamental mode plasmonic resonances, while that in 2.5–5μm is originated from the high-order hybrid mode plasmonic resonances therein. Meanwhile, there does exist a coupled competition effect between the hybrid modes in 2.5–3μm and in 3–5μm. All the results construct the basic guideline for designing these kinds of ISAE materials. Furthermore, we reexamine the physical essence of infrared camouflage based on multispectral bands selective emissivity with thermal management and establish an optimized generalized method for evaluating the camouflage performance of ISAE. The proposed ISAE proves to have much better infrared camouflage property throughout 2.5–14μm than the existing designs reported. The proof-of-principle ISAE is prepared and the selective emission spectrum is characterized, which is in good agreement with the simulations.

3D positioning accuracy and land cover classification performance of multispectral RTK UAVs

Lately, unmanned aerial vehicle (UAV) become a prominent technology in remote sensing studies with the advantage of high-resolution, low-cost, rapidly and periodically achievable three-dimensional (3D) data. UAV enables data capturing in different flight altitudes, imaging geometries, and viewing angles which make detailed monitoring and modelling of target objects possible. Against earlier times, UAVs have been improved by integrating real-time kinematic (RTK) positioning and multispectral (MS) imaging equipment. In this study, positioning accuracy and land cover classification potential of RTK equipped MS UAVs were evaluated by point-based geolocation accuracy analysis and pixel-based ensemble learning algorithms. In positioning accuracy evaluation, ground control points (GCPs), pre-defined by terrestrial global navigation satellite system (GNSS) measurements, were used as the reference data while Random Forest (RF) and Extreme Gradient Boosting (XGBoost) algorithms were applied for land cover classification. In addition, the spectral signatures of some major land classes, achieved by UAV MS bands, were compared with reference terrestrial spectro-radiometer measurements. The results demonstrated that the positioning accuracy of MS RTK UAV is ±1.1 cm in X, ±2.7 cm in Y, and ±5.7 cm in Z as root mean square error (RMSE). In RF and XGBoost pixel-based land cover classification, 13 independent land cover classes were detected with overall accuracies and kappa statistics of 93.14% and 93.37%, 0.92 and 0.93, respectively.

An adaptive image fusion method for Sentinel-2 images and high-resolution images with long-time intervals

Sentinel-2 imagery has garnered significant attention in many earth system studies due to free access and high revisit frequency. Since its spatial resolution is insufficient for many applications, e.g., fine-grained land cover mapping, some studies employ fusion technique that combines high-resolution RGB images with Sentinel-2 multispectral images to improve the resolution of the latter. However, there are two issues in the existing image fusion methods. First, these methods usually assume that the time intervals between images are short (within several days), which is a strong assumption for large-scale high-resolution images and many real-world applications. Second, the spectral discrepancy between multispectral and RGB images could induce spectral aberrations in Sentinel-2 imagery upon fusion. To alleviate these issues, we propose an adaptive image fusion approach named S2IFNet, adaptively fusing images with long-time intervals (from months to years) and spectral inconsistency, thereby increasing the multispectral band resolution of Sentinel-2 imagery. Building on top of the feature extraction and fusion modules, we propose a spectral feature compensation module and a change-aware feature reconstruction module. The former alleviates the possible degradation of spectral attributes in Sentinel-2 imagery resulting from feature fusion. The latter integrates semantic and texture information to avoid adding fake textures caused by land cover changes over time. The experiments demonstrate that S2IFNet surpasses existing image fusion and reference-based super-resolution methods on synthetic and real datasets, yielding fusion results that are clearer and more reliable.

Estimation of Wheat Yield on a Farm in Najaf/Iraq, using Principal Component Analysis of Multi Temporal Satellite Images Pairs

The present work aims to estimate the yield of wheat crop for a particular wheat farm in Najaf Governorate, southwestern Iraq. Seven multispectral bands of the Landsat satellite (8 and 9) were used for two different time periods. The satellite image for the 1st time (Landsat 8) represents the high vegetative growth of the wheat crop, and the satellite image for the second time (Landsat 9) represents the farm after harvest. Changes in land use were detected based on Principal Component Analysis (PCA) technology wherein PCA images were used to calculate wheat production for a specific area in Najaf scene in Iraq. RGB color model was adopted as an unsupervised method of scene classification as this model was used to determine the number of classes in a scene. Maximum-likelihood method was applied as supervised classification with the images generated by applying the principal component analysis technique for the 1st time and for the 2nd time separately considering the number of classes derived from the RGB color model. The data was read within the borders of the region of interest (a wheat farm) for which the yield will be calculated by creating a mask in which the outer borders of the study area are defined. The results showed that the rate of wheat production for the study area amounted to 97.85 tons, with an error rate that did not exceed (1.55%) when using PC images.

Razvoj normaliziranog indeksa tla za urbane studije upotrebom podataka daljinskih mjerenja

This paper presents two novel spectral soil area indices to identify bare soil area and distinguish it more accurately from the urban impervious surface area (ISA). This study designs these indices based on medium spatial resolution remote sensing data from Landsat 8 OLI dataset. Extracting bare soil or urban ISA is more challenging than extracting water bodies or vegetation in multispectral Remote Sensing (RS). Bare soil and the urban ISA area often were mixed because of their spectral similarity in multispectral sensors. This study proposes Normalized Soil Area Index 1 (NSAI1) and Normalized Soil Area Index 2 (NSAI2) using typical multispectral bands. Experiments show that these two indices have an overall accuracy of around 90%. The spectral similarity index (SDI) shows these two indices have higher separability between soil area and ISA than previous indices. The result shows that percentile thresholds can effectively classify bare soil areas from the background. The combined use of both indices measured the soil area of the study area over 71 km2. Most importantly, proposed soil indices can refine urban ISA measurement accuracy in spatiotemporal studies.

Machine Learning Classifier Evaluation for Different Input Combinations: A Case Study with Landsat 9 and Sentinel-2 Data

High-resolution multispectral remote sensing images offer valuable information about various land features, providing essential details and spatially accurate representations. In the complex urban environment, classification accuracy is not often adequate using the complete original multispectral bands for practical applications. To improve the classification accuracy of multispectral images, band reduction techniques are used, which can be categorized into feature extraction and feature selection techniques. The present study examined the use of multispectral satellite bands, spectral indices (including Normalized Difference Built-up Index, Normalized Difference Vegetation Index, and Normalized Difference Water Index) for feature extraction, and the principal component analysis technique for feature selection. These methods were analyzed both independently and in combination for the classification of multiple land use and land cover features. The classification was performed for Landsat 9 and Sentinel-2 satellite images in Delhi, India, using six machine learning techniques: Classification and Regression Tree, Minimum Distance, Naive Bayes, Random Forest, Gradient Tree Boosting, and Support Vector Machine on Google Earth Engine platform. The performance of the classifiers was evaluated quantitatively and qualitatively to analyze the classification results with whole image (comprehensive feature) and small subset (targeted feature). The RF and GTB classifiers were found to outperform all others in the quantitative analysis of all input combinations for both Landsat 9 and Sentinel-2 datasets. RF achieved a classification total accuracy of 96.19% for Landsat and 96.95% for Sentinel-2, whereas GTB achieved 91.62% for Landsat and 92.89% for Sentinel-2 in all band combinations. Furthermore, the RF classifier achieved the highest F1 score of 0.97 in both the Landsat and Sentinel datasets. The qualitative analysis revealed that the PCA bands were particularly useful to classifiers in distinguishing even the slightest differences among the feature class. The findings contribute to the understanding of feature extraction and selection techniques for land use and land cover classification, offering insights into their effectiveness in different scenarios.

A Proposal for Automatic Coastline Extraction from Landsat 8 OLI Images Combining Modified Optimum Index Factor (MOIF) and K-Means

The coastal environment is a natural and economic resource of extraordinary value, but it is constantly modifying and susceptible to climate change, human activities and natural hazards. Remote sensing techniques have proved to be excellent for coastal area monitoring, but the main issue is to detect the borderline between water bodies (ocean, sea, lake or river) and land. This research aims to define a rapid and accurate methodological approach, based on the k-means algorithm, to classify the remotely sensed images in an unsupervised way to distinguish water body pixels and detect coastline. Landsat 8 Operational Land Imager (OLI) multispectral satellite images were considered. The proposal requires applying the k-means algorithm only to the most appropriate multispectral bands, rather than using the entire dataset. In fact, by using only suitable bands to detect the differences between water and no-water (vegetation and bare soil), more accurate results were obtained. For this scope, a new index based on the optimum index factor (OIF) was applied to identify the three best-performing bands for the purpose. The direct comparison between the automatically extracted coastline and the manually digitized one was used to evaluate the product accuracy. The results were very satisfactory and the combination involving bands B2 (blue), B5 (near infrared), and B6 (short-wave infrared-1) provided the best performance.

Assessments of the impacts of environmental factors on vegetation cover at gas flaring sites in the Niger Delta, Nigeria

This study aims at the assessment of the effects of environmental factors on vegetation cover at the flaring sites in Rivers State, Nigeria. Twenty one Landsat 7 Enhanced Thematic Mapper Plus (ETM) data, and four Landsat 8 Operational Land Imager and Thermal Infrared Sensor (OLI-TIRS) data dated from 21/04/2000 to 05/02/2022 with 3 cloud cover were employed to study eleven gas flaring sites. Matrix Laboratory MATLAB code was developed for data processing and analysis. Normalized Differential Vegetation Index (NDVI) was computed from the atmospherically corrected multispectral bands (1-4) for Landsat 7, and bands (2-5) for Landsat 8. Change in NDVI was computed as NDVI450-60)m in the N, E, S and W directions which is the difference between NDVI at 450 m and 60 m distance from the flare stack. The available environmental factors (Facility size, flare stack height and time i.e. year, month and day) were applied to the (NDVI450-60m) in the 4 cardinal directions. Pairwise linear and multiple regression statistical analyses were adopted to investigate the relationships between each of the (NDVI450-60)mN, (NDVI450-60)mE, (NDVI450-60)mS and (NDVI450-60)mW and facility size, stack height and time. The results show that only 12 of the variance in (NDVI450-60)mN is explained by the available data. Therefore, it can be concluded that the combination of the facility size, flare stack height and time accounted for only 12 of the results. Further researches using the rate and volume of flared gas data, and the vegetation species is required to achieve better results.

Unmanned Aerial Vehicle (UAV) and Spectral Datasets in South Africa for Precision Agriculture

Remote sensing data play a crucial role in precision agriculture and natural resource monitoring. The use of unmanned aerial vehicles (UAVs) can provide solutions to challenges faced by farmers and natural resource managers due to its high spatial resolution and flexibility compared to satellite remote sensing. This paper presents UAV and spectral datasets collected from different provinces in South Africa, covering different crops at the farm level as well as natural resources. UAV datasets consist of five multispectral bands corrected for atmospheric effects using the PIX4D mapper software to produce surface reflectance images. The spectral datasets are filtered using a Savitzky–Golay filter, corrected for Multiplicative Scatter Correction (MSC). The first and second derivatives and the Continuous Wavelet Transform (CWT) spectra are also calculated. These datasets can provide baseline information for developing solutions for precision agriculture and natural resource challenges. For example, UAV and spectral data of different crop fields captured at spatial and temporal resolutions can contribute towards calibrating satellite images, thus improving the accuracy of the derived satellite products.

Training sample selection for robust multi-year within-season crop classification using machine learning

Within-season crop classification using multispectral imagery is an effective way to generate timely crop maps that can support water and crop management; however, developing such models is challenging due to limited satellite imagery and ground truth data available during the season. This study investigated ways to optimize the use of multi-year samples in a within-season crop classification model, aiming to enable accurate within-season crop mapping across years. Our study focused on classifying field-scale corn/maize, cotton, and rice in south-eastern Australia from 2013 to 2019. The crop classification model was based on the random forest and support vector machine algorithms applied to Landsat 8 multispectral bands. We designed four experiments to understand the influences of training sample selection on model accuracy. Specifically, we analyzed how the within-season classification accuracies are affected by 1) training sample size; 2) proportions of classification classes; 3) the inclusion of a non-crop class (e.g., fallow land) in the training sample, and 4) training samples collected from different years. We found that 1) the training sample size should be sufficiently large to ensure within-season classification accuracy; 2) using training samples for each crop type in proportion to their occurrence within the landscape results in more accurate multi-year classification; 3) the inclusion of the non-crop class can reduce the accuracy with which crop types are distinguished, so the proportion of the non-crop class should be maintained at a relatively low level, and 4) predicting the current year with training samples from previous years can lead to a minor decline in accuracy compared to using samples only from the current year. These training sample settings were adopted to develop a final model. We found that the model accuracy continues to improve as more input imagery is added as the cropping season progresses, with a rapid rate of initial improvement which then slows. December, the third month of the summer growing season, is the earliest time that reliable maps were generated, with an overall accuracy of 86 % and user’s accuracies for all crops exceeding 80 %. Our proposed experiments are robust and transferable to other regions and seasons to assist the development of within-season crop maps, and can thus be valuable tools to support agricultural management.

A Multispectral UAV Imagery Dataset of Wheat, Soybean and Barley Crops in East Kazakhstan

This study introduces a dataset of crop imagery captured during the 2022 growing season in the Eastern Kazakhstan region. The images were acquired using a multispectral camera mounted on an unmanned aerial vehicle (DJI Phantom 4). The agricultural land, encompassing 27 hectares and cultivated with wheat, barley, and soybean, was subjected to five aerial multispectral photography sessions throughout the growing season. This facilitated thorough monitoring of the most important phenological stages of crop development in the experimental design, which consisted of 27 plots, each covering one hectare. The collected imagery underwent enhancement and expansion, integrating a sixth band that embodies the normalized difference vegetation index (NDVI) values in conjunction with the original five multispectral bands (Blue, Green, Red, Red Edge, and Near Infrared Red). This amplification enables a more effective evaluation of vegetation health and growth, rendering the enriched dataset a valuable resource for the progression and validation of crop monitoring and yield prediction models, as well as for the exploration of precision agriculture methodologies.

Semantic Segmentation with Multispectral Satellite Images of Waterfowl Habitat


 
 
 Migratory waterfowl (i.e., ducks, geese, and swans) management relies on landscape bioenergetic models to inform on-the-ground habitat conditions and conservation practices. Therefore, conservation planners rely on accurate predictions of wetland habitats for waterfowl at regional scales. Unharvested flooded corn is a popular management tool on public and private lands that greatly increases landscape-level energy compared to other wetlands; thus, landscape bioenergetic models are particularly sensitive to these habitat features. Despite their importance to conservation planning and implementation, the abundance and distribution of unharvested flooded corn fields across North America is unknown. Furthermore, training data is difficult to collect and accurate predictions are challenging given their unique attributes and discreteness at landscape-level lens. Advances in multispectral imagery and deep learning algorithms may enable continuous and autonomous detection of these habitat features. Therefore, we conducted modeling experiments using training data of unharvested flooded corn fields in West Tennessee and multispectral imagery collected from Sentinel-2 satellite missions. We performed several experiments using individual band combination composites and/or vegetation indices to identify optimal bands using MRUNET architectures. We subsequently used 3 ensemble models of important individual networks. We found the use of multispectral bands was necessary and although the CIR composite and OSAVI index improved precision, the 12-band composite increased recall, the metric we were most interested in. Moreover, all ensembles exhibited poor performance. Here, we present results of our initial modeling experiments and suggest future modeling exercises including temporal image and vegetation index stacking using multi-modal and/or recurrent neural network architectures.
 
 

Enhancing Soil Mapping with Hyperspectral Subsurface Images generated from soil lab Vis-SWIR spectra tested in southern Brazil

The Remote Sensing (RS) products involving the visible, near-infrared and short-wave infrared (Vis–NIR-SWIR, simply Vis-SWIR) wavelengths are applicable for spatial prediction of soil properties, usually in the process called Digital Soil Mapping (DSM). The multispectral optical orbital sensors lack spectral resolution, while the Proximal Soil Sensing (PSS) offers many application possibilities. The limited spectral resolution and soil cover are the main challenging aspects for DSM. The objective of the study was to combine the soil lab Vis-SWIR spectra within DSM technique to create new hyperspectral images to use as covariates in a novel method of soil mapping. The advantages of the method were demonstrated throughout the spatial prediction of the Total Carbon (TC) content on soils of the upper part of Itatiaia National Park (INP), (Rio de Janeiro and Minas Gerais States, southeastern Brazil). The covariates applied were multispectral orbital bands from Sentinel 2, relief with its terrain derivations, geology and geomorphology. Using uppermost soil horizon lab Vis-SWIR spectra information from 72 points in the upper part of INP, 130 hyperspectral images were generated, in other words, Hyperspectral Subsurface Image. The validation methods were 8-fold cross-validation (CV) and External Validation (EV) running alongside within Random Forest algorithm, where randomness was locked. The DSM reaches an average CV coefficient of determination (R2) of 0.39 and Root Mean Square Error (RMSE) of 4.6, while creating a new model with a set of hyperspectral images gave a CV R2 of 0.60 and RMSE of 4.06. The EV also presented an advantage for the second approach. The results support the method as the first strong integration between PSS and DSM, named by the authors as Hyperspectral Soil Mapping (HSM), which proved to be more efficient for mapping TC due to the pure soil lab Vis-SWIR spectra signal (free of atmospheric influence and vegetation cover). The method can be applied in dense or sparsely vegetated areas, for agricultural or conservation purposes, for various soil properties, with each and every wavelength.

Predicting Maize Theoretical Methane Yield in Combination with Ground and UAV Remote Data Using Machine Learning.

The accurate, timely, and non-destructive estimation of maize total-above ground biomass (TAB) and theoretical biochemical methane potential (TBMP) under different phenological stages is a substantial part of agricultural remote sensing. The assimilation of UAV and machine learning (ML) data may be successfully applied in predicting maize TAB and TBMP; however, in the Nordic-Baltic region, these technologies are not fully exploited. Therefore, in this study, during the maize growing period, we tracked unmanned aerial vehicle (UAV) based multispectral bands (blue, red, green, red edge, and infrared) at the main phenological stages. In the next step, we calculated UAV-based vegetation indices, which were combined with field measurements and different ML models, including generalized linear, random forest, as well as support vector machines. The results showed that the best ML predictions were obtained during the maize blister (R2)-Dough (R4) growth period when the prediction models managed to explain 88-95% of TAB and 88-97% TBMP variation. However, for the practical usage of farmers, the earliest suitable timing for adequate TAB and TBMP prediction in the Nordic-Baltic area is stage V7-V10. We conclude that UAV techniques in combination with ML models were successfully applied for maize TAB and TBMP estimation, but similar research should be continued for further improvements.

A spectral preserved model based on spectral contribution and dependence with detail injection for pansharpening

Pansharpening integrates the high spectral content of multispectral (MS) images and the fine spatial information of the corresponding panchromatic (PAN) images to produce a high spectral-spatial resolution image. Traditional pansharpening methods compensate for the spatial lack of the MS image using the PAN image details, which easily causes spectral distortion. To achieve spectral fidelity, a spectral preservation model based on spectral contribution and dempendence with detail injection for pansharpening is proposed. In the proposed model, first, an efficacy coefficient (CE) based on the spatial difference between the MS and PAN images is designed to suppress the impact of the detail injection on the spectra. Second, the spectral contribution and dependence (SCD) between the MS bands and pixels are considered to strengthen the internal adaptation of the spectra. Finally, a spectrally preserved model based on CE and SCD is designed to force the fused image fidelity in spectra when the MS image is pansharpened with the details of the PAN image. Experimental results show that the proposed model is effective.

Visual Detection and Association Tracking of Dim Small Ship Targets from Optical Image Sequences of Geostationary Satellite Using Multispectral Radiation Characteristics

By virtue of the merits of wide swath, persistent observation, and rapid operational response, geostationary remote sensing satellites (e.g., GF-4) show tremendous potential for sea target system surveillance and situational awareness. However, ships in such images appear as dim small targets and may be affected by clutter, reef islands, clouds, and other interferences, which makes the task of ship detection and tracking intractable. Considering the differences in visual saliency characteristics across multispectral bands between ships and jamming targets, a novel approach to visual detecting and association tracking of dense ships based on the GF-4 image sequences is proposed in this paper. First, candidate ship blobs are segmented in each single-spectral image of each frame through a multi-vision salient features fusion strategy, to obtain the centroid position, size, and corresponding spectral grayscale information of suspected ships. Due to the displacement of moving ships across multispectral images of each frame, multispectral association with regard to the positions of ship blobs is then performed to determine the final ship detections. Afterwards, precise position correction of detected ships is implemented for each frame in image sequences via multimodal data association between GF-4 detections and automatic identification system data. Last, an improved multiple hypotheses tracking algorithm with multispectral radiation and size characteristics is put forward to track ships across multi-frame corrected detections and estimate ships’ motion states. Experiment results demonstrate that our method can effectively detect and track ships in GF-4 remote sensing image sequences with high precision and recall rate, yielding state-of-the-art performance.