Thermal Infrared Images Research Articles

Estimation of Evapotranspiration from the People’s Victory Irrigation District Based on the Data Mining Sharpener Model

Reasonable evaluation of evapotranspiration (ET) is crucial for optimizing agricultural water resource management. In the study, we utilized the Data Mining Sharpener (DMS) model; the Landsat thermal infrared images were sharpened from a spatial resolution of 100 m to 30 m. We then used the Surface Energy Balance System (SEBS) to estimate daily ET during the winter wheat growing season in the People’s Victory Irrigation District in Henan, China. It was concluded that the spatiotemporal patterns of land surface temperature and daily evapotranspiration remained consistent before and after sharpening. Results showed that the R2 value between the ET of 30 m spatial resolution and the value by eddy covariance method reached 0.814, with an RMSE of 0.516 mm and an MAE of 0.245 mm. All of these were higher than those of 100 m spatial resolution (R2 was 0.802, the RMSE was 0.534 mm, and the MAE was 0.253 mm). Furthermore, the daily ET image with a 30 m spatial resolution exhibited clear texture and distinct boundaries, without any noticeable mosaic effects. The changes in surface temperature and ET were more consistent in complex subsurface environments. The daily evapotranspiration of winter wheat was significantly higher in areas with intricate drainage systems compared to other regions. During the early growth stage, daily evapotranspiration decreased steadily until the overwintering stage. After the greening and jointing stages, it began to increase and peaked during the sizing period. The correlation between net solar radiation and temperature with ET was significant, while relative humidity and soil moisture were negatively correlated with ET. Throughout the growth period, net solar radiation had the greatest effect on ET.

Infrared Thermal Imaging Analysis in Screening for Toddler’s Fracture: A Proof-of-Concept Study

This study explored and developed high-resolution infrared thermal (HRIT) imaging for screening toddler’s fractures. A toddler’s fracture is a common tibial fracture in children younger than six years old. The study included 39 participants admitted to an emergency department with a suspected toddler’s fracture. X-ray confirmed eight participants with a toddler’s fracture (20.5%). Infrared images of participants were recorded on their index visit, focusing on region-of-interests on the injured and the contralateral (uninjured) legs. The uninjured leg acted as a thermal reference. Six statistical measures obtained from the images were analyzed. These were maximum, mean, standard deviation, median, interquartile range, and skewness. The Shapiro–Wilk test indicated that the measures were from a normal distribution. A two-sample t-test indicated that the majority of the six measures had significantly different means (p < 0.05) when comparing the participants with and without a fracture. Similarly, the first principal component (PC1), obtained through principal component analysis of the six measures, was significantly different (p < 0.05) comparing participants with and without a fracture. Visualization of the statistical measures and their PC1 demonstrated distinct clustering. This study demonstrated that HRIT imaging is valuable for screening for toddler’s fractures, but a larger follow-on study will be required to confirm the findings.

Estimating rectal temperature in exercising horses in a competition environment using infrared thermal imaging

Abstract Short-term intense and or prolonged medium intensity exercise in conditions of moderate-high environmental temperature or temperature and humidity can result in moderate to marked hyperthermia. This in turn may result in decreased exercise capacity, earlier onset of fatigue and increased risk of falls or injury. Early identification of moderate-severe hyperthermia is important to allow appropriate intervention, either before, during or after exercise. Infra-red thermal imaging (IRT) allows measurement of horse surface temperature non-invasively, however it is generally considered to be unreliable in outdoor environments. Collection of pilot data suggested that this modality warranted further investigation. IRT Images were obtained from horses during training and competition at the 2019 Ready Steady Tokyo Olympic equestrian test event (Wet Bulb Globe Temperature [WBGT] > 28 °C), during and after the cross-country at the 2019 Burghley Horse Trials CCI 5* (WBGT 12-15 °C), the 2019 Adelaide International Horse Trials CCI 4* (WBGT ∼ 20 °C) and during the 2020 Tokyo Olympic Games (held in 2021; WBGT > 28 °C). A total of 87 paired hindquarter IRT images and rectal temperature measurements were obtained. When all data from all 4 events were pooled the bias and limits of agreement were −0.3 ± 1.1 °C (n = 87), i.e. IRT underestimated rectal temperature. When the data from Ready Steady Tokyo 2019 and Tokyo 2020 were combined as both events took place in similar thermal environmental conditions) the bias and limits of agreement were −0.1 ± 0.8 °C (n = 45). For 29 paired hindquarter thermal images, where a second image was obtained within 15 s, there was no significant difference between the mean of the first and second images (Image 1: 38.9 ± 0.9 °C versus Image 2: 38.7 ± 0.8 °C; ; n = 29). In conclusion, IRT imaging in environmental conditions of high temperature or high temperature and humidity may be useful in identifying horses with moderate to severe hyperthermia.

A deep learning approach combining DeepLabV3+ and improved YOLOv5 to detect dairy cow mastitis

Dairy cow mastitis has a great impact on the productivity of dairy cows and the profits of livestock farms. Early detection is of great significance to improve the efficiency of mastitis treatment. However, due to the low resolution of thermal infrared images and the complexity of the living environment of dairy cows, it is difficult to detect the eyes and udders of cows, which reduces the detection accuracy of mastitis. To solve this problem, this paper proposes a two-stage model (DCYOLO) integrating the DeepLabV3 + semantic segmentation network and an improved YOLOv5 target recognition network, which is used to detect the eyes and udders of dairy cows under complex background and applied to the severity classification of dairy cow mastitis. In the first stage, the DeepLabV3 + model was used to segment the eyes and udders of dairy cows from thermal infrared images. In the second stage, the segmented image was input into the target recognition YOLOv5 network for key parts recognition. Finally, to further improve the detection accuracy of the model, a convolutional block attention module (CBAM) was added at the end of the main part of the YOLOv5 model. After comparing different semantic segmentation and target recognition networks, the DeepLabV3 + network and YOLOv5 network performed best. The mIoU and mean pixel accuracy (mPA) of the DeepLabV3 + network reached 86.98 % and 92.92 %, respectively. The mean average precision (mAP) and F1 scores of the YOLOv5 network for unsegmented thermal infrared images reached 93.4 % and 90.9 %, respectively. The CBAM-added YOLOv5 (CAYOLOv5) model was combined with the DeepLabV3 + model. Compared with the single YOLOv5 model, the mAP and F1 scores of DCYOLO increased by 5.4 % and 5.3 %, respectively. Therefore, the proposed model can achieve more accurate positioning of key parts of dairy cows. Based on this model, the eye and udder temperature differences of 50 dairy cows were extracted for mastitis detection, and the detection results were compared with the results of the somatic cell count (SCC) approach. The results showed that the classification accuracy of mastitis was 86 %, and the average sensitivity and specificity were 79.41 % and 92.49 %, respectively. The dairy cow mastitis detection method based on the two-stage model can accurately locate the key parts of dairy cows and realize the automatic detection and classification of dairy cow mastitis, and the accuracy is high.

AN IMPROVED IMAGE REGISTRATION ALGORITHM FOR THERMAL INFRARED AND PANCHROMATIC IMAGE BASED ON GEOMETRIC STRUCTURAL PROPERTIES

Abstract. In order to improve the image registration accuracy of heterogeneous remote sensing images with large radiation differences, an improved image registration algorithm for thermal infrared and panchromatic images is proposed. This method uses the phase consistency of intensity and direction to construct a geometric structure feature descriptor called the Histogram of Oriented Phase Congruency (HOPC) for image registration. It employs the Curvature Scale Space (CSS) corner detection to concentrate and extract feature points in contour areas, and defines a similarity metric (called HOPCn) based on the Euclidean distance between the descriptors for high-precision heterogeneous images registration. Five sets of panchromatic and thermal infrared images were selected for verification. The results show that the improved algorithm can match the homonymy points around the image contour, and the Correct Matching Rate (CMR) is improved by 2.7% and the Root Mean Square Error (RMSE) is improved by 3.7% compared to the original algorithm. This proves that the proposed method has good robustness under conditions of large radiation differences.

SGFNet: Semantic-Guided Fusion Network for RGB-Thermal Semantic Segmentation

Recently, semantic segmentation based on RGB and thermal infrared (TIR) images has become a research hotspot because of its stability in the weak light environment. However, most of the current methods ignore the differences between the two modalities of data and do not use semantic information in multi-modal fusion. In this paper, we propose a novel Semantic-Guided Fusion Network (SGFNet) for RGB-Thermal semantic segmentation, which makes full use of semantic information in the multi-modal fusion. Our SGFNet consists of an asymmetric encoder with TIR branch and RGB branch and a decoder. We concentrate on enhancing the multi-modal feature representation in the encoder with a pattern of fusion and enhancement. Specifically, considering that TIR images are stable under weak light conditions, we first propose a Semantic Guidance Head to extract semantic information in the TIR branch. In the RGB branch, we propose a Multi-modal Coordination and Distillation Unit to fuse multi-modal features first. Then, we propose a Cross-level and Semantic-guided Enhancement Unit to enhance the fused features with cross-level information and semantic information. We arrange these two units at all stages of the RGB branch to generate features with strong representation abilities at different levels. For the decoder, to obtain large receptive fields and fine edges, we improve the Lawin ASPP decoder by introducing edge information extracted from the low-level features, proposing the edge-aware Lawin ASPP decoder. With our encoder and decoder working together, our SGFNet can identify objects accurately and segment objects finely. Extensive experiments on the MFNet dataset demonstrate the superior performance of the proposed SGFNet compared with state-of-the-art methods. The code and results of our method are available at https://github.com/kw717/SGFNet.

Mask R-CNN aided fruit surface temperature monitoring algorithm with edge compute enabled internet of things system for automated apple heat stress management

Our prior study focused on development of internet of things (IoT) and edge-compute enabled crop physiology sensing system (CPSS) for apple sunburn monitoring. Edge compute algorithm on CPSS estimated sunburn susceptibility as fruit surface temperature (FST) through pixel-by-pixel multiplication of captured thermal infrared images with segmented fruits binary mask. The segmentation was performed using color-based K means clustering approach. This limited CPSS applicability to monitor sunburn of red colored cultivars only and when fruits develop color, typically late growing season. This is a key research gap as recent weather patterns have shown that sunburn can occur during early growing season when fruits are green to yellow. Therefore, aim of this study was to develop and field evaluate cultivar and color independent mask region-convolution neural network (R-CNN) aided fruit segmentation model and edge compute compatible FST estimation algorithm. Season long field data were collected in 2021 using eight CPSS nodes (three in cv. WA38 [Cosmic crisp] and five in cv. Honeycrisp). Collected data were used to develop and validate mask R-CNN based fruit segmentation model. Developed mask R-CNN based model was able to segment fruits of two apple cultivars and of varying colors with 91.4 % average precision. In orchard evaluations (2022 season), the resulting algorithm ported on CPSS was able to accurately segment (dice similarity coefficient = 0.89) and estimate apple FST with < 0.5 °C error compared to ground truth data. With compute time of about 37 s, data processing time was reduced by 22 % over previous algorithm. High ambient temperature (>35 °C) on a warmer day resulted in multiple throttling errors caused by excessive CPU temperature; however, the CPSS performance was uncompromised in FST estimation. Ambient air temperature did not affect RAM utilization and CPU clock frequency. Overall, developed FST algorithm can potentially be used as input to actuate water-based cooling system.

RETRACTED ARTICLE: Dual Siamese Anchor Points Adaptive Tracker with Transformer for RGBT Tracking

Due to environmental conditions, such as rainy days, foggy days, and dim lighting, objects in visible light images are not prominently displayed, leading to an easy loss of targets during tracking. In recent years, many RGB visible light trackers have achieved significant success in addressing visual tracking challenges. However, these trackers perform poorly when tracking targets under special conditions, such as occlusions and low-light scenarios. In contrast, objects in thermal infrared images are more distinct in poor lighting conditions. Given this characteristic, researchers have shown increased interest in the development of trackers that combine thermal infrared and visible light imagery. However, some mainstream RGBT (red–green–blue and thermal) algorithms, such as MANET and ADNET, are based on the anchor-based theory, requiring consideration of anchor box sizes and introducing a substantial number of hyperparameters. This can lead to suboptimal performance when tracking dynamically changing targets. Moreover, these models rely on convolutional neural networks for feature extraction, which have limitations in capturing global features. In this paper, we introduce a novel training network model called DAPAT, which combines the anchor-free concept with Transformer theory. DAPAT differs from previous models in several ways. Specifically, we have designed a straightforward model to extract precise global features from template and search images. We have also incorporated two enhancement modules into the model to improve template and search images of different sizes while suppressing the impact of non-target images. We employ a dual-stream feature fusion network to reduce the loss of image feature information due to feature correlation operations. Finally, we compare the performance of the tracking model proposed in this paper with some advanced RGBT trackers on three data sets (RGBT234, RGBT210, and GTOT). The test results demonstrate that our tracker exhibits improvements in robustness and success rate, among other performance aspects.

Evaluation of spatial downscaling for satellite retrieval of evapotranspiration from the nonparametric approach in an arid area

Evapotranspiration (ET) data is critical for monitoring the limited hydrological resources in arid areas. At present, the spatial resolution of ET derived by the thermal infrared images from satellite data is coarse (always more than 500 m). The low-resolution data cannot meet the needs of the high-precision mapping of ET. Therefore, it is important to undertake a downscaling procedure to improve the spatial resolution of ET. This paper selects the oasis region and surrounding deserts of Zhangye City of China as study area and compares with two downscaling procedures that include input and output downscaling procedure. The ET is estimated with the Remote Sensing-Non Parametric (RS-NP) model. The two downscaling procedures use Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM) to improve the spatial resolution of ET from 1-km to 100 m. Compared with the result derived from the output downscaling procedure, the high-resolution ET derived from the input downscaling procedure has a slightly better performance with R2 of 0.92 and RMSE of 41.8 W/m2, respectively. Neglectingthe downscaling of land surface temperature and broadband reflectance (BBR) result in larger errors of ET estimation with bias values of 73.0 W/m2 and 23.7 W/m2, respectively. Those values account for almost 1/3 and 1/10 of ET values, respectively. In non-vegetated areas, those proportions increase further around 90.6 % and 21.3 %, respectively. Furthermore, the influence of the neglect of the downscaling of the near-infrared reflectance on the ET estimation dominates that of BBR with a proportion of 42 %. The input downscaling procedure is suitable for generating the high-precision map of ET, especially in vegetated areas and during northern hemisphere summer months. Our study is helpful to obtain ET maps in fine spatial resolutions and contribute to the comprehensive understanding of the importance of the downscaling of each kind of input for reliable ET estimations.

Combined Retrieval of Oil Film Thickness Using Hyperspectral and Thermal Infrared Remote Sensing

An outdoor experiment was conducted to measure the thickness of oil films (0~3000 μm) using an airborne hyperspectral imager and thermal infrared imager, and the spectral response and thermal response of oil films of different thicknesses were analyzed. The classic support vector regression (SVR) model was used to retrieve the oil film thickness. On this basis, the suitable range for retrieving oil film thickness using hyperspectral and thermal infrared remote sensing was explored, and the decision-level fusion algorithm was developed to fuse the retrieval capabilities of hyperspectral and thermal infrared remote sensing for oil film thickness. The following conclusions can be drawn: (1) Based on airborne hyperspectral data and thermal infrared data, the retrieval accuracy of oil films of different thicknesses reached 154.31 μm and 116.59 μm, respectively. (2) The S185 hyperspectral data were beneficial for retrieving thicknesses greater than or equal to 400 μm, and the H20T thermal infrared data were beneficial for retrieving thicknesses greater than 500 μm. (3) The result of the decision-level fusion model based on a fuzzy membership degree was superior to those obtained using a single sensor (hyperspectral or thermal infrared), indicating that it can better integrate the retrieval results of hyperspectral and thermal infrared remote sensing for oil film thickness. Furthermore, the feasibility of using hyperspectral and thermal infrared remote sensing to detect water-in-oil emulsions of different thicknesses was investigated through spectral response and thermal response analysis.

Characterizing SO2 Emission Rate, Thermal Anomalies, from Opened and Closed Vent System at Agung, Bromo, and Sinabung Volcanoes in Indonesia

Agung, Bromo, and Sinabung Volcanoes have high volcanic activity over the last decade, and have different eruption characteristics. Hence, it would be fascinating to study the characteristics of their volcanic activity patterns based on SO2 emission rates and thermal anomaly correlated with the seismicity data. The SO2 emission rate measurement was carried out using the Differential Optical Absorption Spectroscopy (DOAS), and calculated based on SO 2 column density, distance of measurement, wind speed, and wind direction. In addition, SO2 emission was detected using Ozone Monitoring Instrument (OMI) images with daily global coverage. Thermal anomaly detection was performed using Advance Spaceborne Thermal Emission and Reflection Radiometer (ASTER) of Thermal Infrared (TIR) subsystem with high spatial resolution (90x90 m). ASTER TIR images were corrected for radiometric and thermal atmospheric. The emissivity and brightness temperature separation algorithm was applied to obtain surface temperature of Agung, Bromo, and Sinabung Volcanoes. All the data were correlated with the seismicity of each volcano. The SO2 emission rates correlate with the magma ascent to the shallow depth in an open system volcano (Bromo Volcano). In the closed-system volcanoes (early phase of Agung and Sinabung), SO2 emission was detected after the transition of closed to open system. Magmatic injection from the reservoir to the shallow depth was detected as thermal anomalies, such as in Agung Volcano. Whereas in Bromo Volcano, the thermal anomaly was insignificant since Bromo Volcano has an explosive eruption at a short period, so the ASTER image could not observe the thermal anomaly on the eruption time. Thermal anomaly pattern in Sinabung Volcano was the manifestation of new magmatic injection to the shallow depth. Therefore, their increase serves as indicators for the increasing magmatic activity prior to the eruptions. Keywords: SO2 emission rate, thermal anomaly, DOAS, OMI, ASTER, Open Vent, Closed Vent

A RGB-Thermal based adaptive modality learning network for day–night wildfire identification

Wildfires have long been a danger to the atmosphere and ecological environment. With the advancement of deep learning technology and sensor equipment, wildfire identification methods based on optical (RGB) or thermal infrared (TIR) images have made significant progress. However, single-modal identification technologies often experience serious performance degradation in challenging scenarios. Therefore, a multi-modal framework that combines both RGB and TIR is necessary for accurate wildfire identification. Despite this, current RGB-T based identification methods focus solely on developing fusion strategies to learn shared representations of the two modalities, while ignoring the effect of modality-specific features. To address this issue, we propose a novel wildfire identification framework that can adaptively learn modality-specific and shared features. Specifically, our model utilizes two parallel encoders to extract multi-scale RGB and TIR features, and integrates these modality-specific features into the fusion feature layer. Each modality-specific decoder and shared decoder corresponds to three independent label supervisions that oversee feature adaptive learning. To validate the effectiveness of our approach, we propose a new RGB-T paired wildfire semantic segmentation dataset that includes flames of different scales captured under varying lighting conditions from multiple UAV-based and ground-based cameras. Finally, we evaluate our proposed method alongside benchmarking single-modal methods and state-of-the-art multi-modal methods on two tasks: a common test setting under various lighting conditions, and a generalization test across three subsets. The experimental results demonstrate the superiority of our proposed method, with the average ＋6.41% (IoU) and ＋3.39% (F1-score) improvement over the second-best RGB-T semantic segmentation method.

Fabrication of Plastic Optics from Chalcogenide Hybrid Inorganic/Organic Polymers for Infrared Thermal Imaging

AbstractThe development of infrared (IR) plastic optics for infrared thermal imaging, particularly, in the long‐wave IR (LWIR) spectrum (7–14 µm) is an area of growing technological interest due to the potential advantages associated with plastic optics (e.g., moldability and low cost). The development of a new class of optical polymers, chalcogenide‐based inorganic/organic hybrid polymers (CHIPs) derived from the inverse vulcanization of elemental sulfur, has enabled significant improvements in IR transparency due to reduction of IR absorbing organic comonomer units. The vast majority of effort has focused on new chalcogenide hybrid polymer synthesis and optical property improvements (e.g., refractive index, Abbe number, and LWIR transmission); however, fabrication and IR imaging methodology to prepare optical components has not been demonstrated, which remains critical to develop viable IR plastic optics. A new methodology is reported to fabricate optical components and evaluate LWIR imaging performance of this emerging class of optical polymers. New diffractive flat optics with a Fresnel lens design for these materials have been developed, along with a basic LWIR imaging system to evaluate CHIPs for LWIR imaging. This system‐based approach enables correspondence of copolymer structure‐property correlations with LWIR imaging performance, along with demonstration of room temperature LWIR imaging.

Preparation of HfSiO4 via Non-Hydrolytic Sol-Gel method at low temperatures

Hafnon (HfSiO4) can be applied in electronics, dielectric films, refractory materials, and many other critical application areas. To explore an efficient preparation process of HfSiO4, a novel non-hydrolytic sol-gel (NHSG) method was used to synthesize high-purity HfSiO4 in this study. X-ray diffraction (XRD), field emission scanning electron microscopy (SEM), energy-dispersive spectrometer (EDS), thermogravimetric differential scanning calorimetry (TG-DSC), Fourier transform infrared spectroscopy (FT-IR), and infrared thermal imaging (ITI) were used to investigate the synthesis temperature and mechanism of HfSiO4 and the effects of process parameters on its synthesis efficiency. The results show that the prepared sols undergo a highly reactive non-hydrolytic polycondensation reaction at room temperature. High temperature reflux or long time aging step is not necessary to form the strong Hf-O-Si bonds. Furthermore, with the introduction of the mineralizer agent LiF, the synthesis temperature of HfSiO4 can be drastically reduced, and the HfSiO4 with high purity can be synthesized at low temperatures. A moderate molar ratio of silicon to hafnium (Si:Hf) and molar concentration of the solution contribute to improve the synthesis efficiency of HfSiO4, and the purity of HfSiO4 products can reach up to 98.12 %.

BMDENet: Bi-Directional Modality Difference Elimination Network for Few-Shot RGB-T Semantic Segmentation

Few-shot semantic segmentation (FSS) aims to segment the target prospects of query images using a few labeled support samples. Compared with the fully-supervised methods, FSS has a greater ability to generalize to unseen classes and reduce the pressure to label large pixel-level datasets. To cope with the complex outdoor lighting environment, we introduce the thermal infrared images (T) to the FSS task. However, the existing RGB-T FSS methods all ignore the differences between various modalities for direct fusion, which may hinder cross-modal information interaction. Also considering the effect of successive downsampling on the results, we propose a bi-directional modality difference elimination network (BMDENet) to boost the segmentation performance. Concretely, the bi-directional modality difference elimination module (BMDEM) reduces the heterogeneity between RGB and thermal images in the prototype space. The residual attention fusion module (RAFM) mines the bimodal features to fully fuse the cross-modal information. In addition, the mainstay and subsidiary enhancement module (MSEM) enhances the fusion features for the existing problem of the advanced model. Extensive experiments on Tokyo Multi-Spectral-4i dataset prove that BMDENet achieves the state-of-the-art on both 1-and 5-shot settings.

Monitoring Seasonal Fluctuations in Saline Lakes of Tunisia Using Earth Observation Data Processed by GRASS GIS

This study documents the changes in the Land Use/Land Cover (LULC) in the region of saline lakes in north Tunisia, Sahara Desert. Remote sensing data are a valuable data source in monitoring LULC in lacustrine landscapes, because variations in the extent of lakes are visible from space and can be detected on the images. In this study, changes in LULC of the salt pans of Tunisia were evaluated using a series of 12 Landsat 8-9 Operational Land Imager (OLI) and Thermal Infrared (TIRS) images. The images were processed with the Geographic Resources Analysis Support System (GRASS) Geographic Information System (GIS) software. The study area included four salt lakes of north Tunisia in the two regions of the Gulf of Hammamet and Gulf of Gabès: (1) Sebkhet de Sidi el Hani (Sousse Governorate), (2) Sebkha de Moknine (Mahdia Governorate), (3) Sebkhet El Rharra and (4) Sebkhet en Noual (Sfax). A quantitative estimate of the areal extent analysed in this study is 182 km × 185 km for each Landsat scene in two study areas: Gulf of Hammamet and Gulf of Gabès. The images were analysed for the period 2017–2023 on months February, April and July for each year. Spatio-temporal changes in LULC and their climate–environmental driving forces were analysed. The results were interpreted and the highest changes were detected by accuracy assessment, computing the class separability matrices, evaluating the means and standard deviation for each band and plotting the reject probability maps. Multi-temporal changes in LULC classes are reported for each image. The results demonstrated that changes in salt lakes were determined for winter/spring/summer months as detected changes in water/land/salt/sand/vegetation areas. The accuracy of the classified images was evaluated using pixel rejection probability values, which were filtered out using the ‘r.mapcalc’ module of GRASS GIS. The confidence levels were computed and visualised with a series of maps along with the error matrix and measured convergence level of classified pixels. This paper contributes to the environmental monitoring of Tunisian landscapes and analysis of climate effects on LULC in landscapes of north Africa.

A Dual Convolutional Neural Network with Attention Mechanism for Thermal Infrared Image Enhancement

In industrial applications, thermal infrared images, which are commonly used, often suffer from issues such as low contrast and blurred details. Traditional image enhancement algorithms are limited in their effectiveness in improving the visual quality of thermal infrared images due to the specific nature of the application. Therefore, we propose a dual Convolutional Neural Network (CNN) combined with an attention mechanism to address the challenges of enhancing low-quality thermal infrared images and improving their visual quality. Firstly, we employ two parallel sub-networks to extract both global and local features. In one sub-network, we utilize a sparse mechanism incorporating dilated convolutions, while the other sub-network employs Feature Attention (FA) blocks based on channel attention and pixel attention. This architecture significantly enhances the feature extraction capability. The use of attention mechanisms allows the network to filter out irrelevant background information, enabling more flexible feature extraction. Finally, through a simple yet effective fusion block, we thoroughly integrate the extracted features to achieve an optimal fusion strategy, ensuring the highest quality enhancement of the final image. Extensive experiments on benchmark datasets and real images demonstrate that our proposed method outperforms other state-of-the-art models in terms of objective evaluation metrics and subjective assessments. The generated images also exhibit superior visual quality.

CLIM: Co-occurrence with Laplacian Intensity Modulation and Enhanced Color Space Transform for Infrared-Visible Image Fusion

Thermal infrared and multispectral visible remote sensing image fusion combines thermal image information with corresponding visible scene content to generate a better representative fused image. Thermal images can distinguish targets using difference in thermal radiation measurements, whereas visible images contain better texture detail in multispectral wavelength bands. The article presents a novel methodology named CLIM to sharpen coarser spatial resolution multispectral remote sensing images using relatively higher spatial resolution broadband thermal infrared image. The boundary-preserving information is extracted from high resolution thermal infrared image using co-occurrence image filter, and is combined with Laplacian of Gaussian based sharpened image to extract salient features for injection. In addition, visible image is transformed to IHS color space, and intensity component is enhanced using CLAHE and inverse transformation to generate enhanced visible image for fusion. The procedure developed is evaluated with Indian Nano Satellite (INS) broadband thermal infrared images available at a spatial resolution of 175 m with same day acquisition MODIS multispectral visible images available at a relatively coarser spatial resolution of 500 m. The nearest acquisition of Landsat-8 thermal infrared images with MODIS multispectral visible images is also used for infrared–visible multi-modal image fusion. The CLIM fused image confirms that distinct features such as dam, ship docking zones and refinery regions, are better demarked and semantically more meaningful in comparison with individual thermal infrared and multispectral visible image. The proposed CLIM approach is compared with, and found to perform better than state-of-the-art image fusion techniques, both visually and quantitatively.

Sparse mixed attention aggregation network for multimodal images fusion tracking

Recent years have witnessed the exciting performance of trackers based on Transformer. However, they usually separate the process of information extraction and integration, weakening the information interaction between the target and search region. In addition, they depend on traditional Transformer to model the long range dependency, which leads to a lack of focus on the primary information needed by high-accuracy trackers. In this paper, a sparse mixed attention aggregation model is proposed for robust tracking based on visible and thermal infrared images. To be specific, a backbone network composed of sparse mixed attention is designed to achieve information extraction and integration. This is helpful to obtain specific discriminative feature information and enhance their communication. To give full play to the complementary visible and thermal information, a confidence aware aggregation network is designed, which can learn the reliable confidence of visible and thermal branches. Finally, a corner-based localization head is introduced to estimate the target state. Extensive experiments on three large-scale multimodal tracking benchmarks demonstrate the superior tracking ability of the proposed tracker over other advanced trackers.

On-orbit monitoring flying aircraft day and night based on SDGSAT-1 thermal infrared dataset

Aerial moving target detection is an important technology to ensure the safety of civil flight and airspace. Different from ground-based active radar and space-based visible imager, on-orbit thermal infrared sensor (TIS), which records the temperature difference between background and targets, is proved to be a stable means of monitoring the aircraft throughout the day. Limited by low spatial resolution and dataset scarcity, there are few existing the research of space-based aircraft detection based on TIS. In this paper, a new multi-model joint method to detect on-orbit flying aircrafts with only 1 × 1 to 2 × 2 pixels is proposed and validated. And the first three-band thermal infrared flying aircrafts dataset (TIFAD) with 30 m spacial resolution is produced from SDGSAT-1 thermal infrared images, including the analysis of its infrared feature. Since the point target cannot by distinguished by the shape, all samples are selected by an improved aircraft contrails detection algorithm based on progressive probabilistic Hough transform, from more than 325GB SDGSAT-1 TIS remote sensing images (RSIs). And training by TIFAD, state-of-the-art YOLOv8 algorithm is applied to more than 17,767 medium spatial resolution thermal infrared images, and the potential aircrafts (with contrails or non-contrails) in RSIs are effectively detected. Moreover, the giga floating point operations (GFLOPs) of detecting flying targets from RSIs with 300 km width range is only 8.9. This paper is believed to provide a technical support of detecting aircraft targets around the globe throughout the day, and supplement on-orbit flying targets dataset of TIS images.