Search

The key to hyperspectral image (HSI) anomaly detection is effectively distinguishing anomalies from the background. Various detectors, such as low-rank sparse representation and background estimation, have been proposed. However, they usually exhibit poor performance due to complex backgrounds and similar anomalies and backgrounds. Considering that most of the pixels in HSI are background, the autoencoder can emphatically and automatically extract and reconstruct background features. And background suppression can further enhance the spectral differentiation of the anomalies and backgrounds. Therefore, we propose a new anomaly detector by integrating autoencoder and background suppression. First, the anomaly coefficient of HSI data is extracted via the local characteristics. The HSI data is fed into the 2Dimensional (2D) convolution autoencoder to reconstruct the background. Since the reconstruction background contains incorrect anomalies, the anomaly coefficient isused to suppress the reconstruction image. Then, the detection map is obtained with the Mahalanobis distance. The experiment is performed in different HSI datasets. The results illustrate that the proposed method is effective and outperforms other advanced methods in terms of detection performance. [ABSTRACT FROM AUTHOR]

Purpose: Background suppression (BS) is recommended in arterial spin labeling (ASL) for improved SNR but is difficult to optimize in existing velocity-selective ASL (VSASL) methods. Dual-module VSASL (dm-VSASL) enables delay-insensitive, robust, and SNR-efficient perfusion imaging, while allowing efficient BS, but its optimization has yet to be thoroughly investigated. Methods: The inversion effects of the velocity-selective labeling pulses, such as velocity-selective inversion (VSI), can be used for BS, andwere modeled for optimizing BS in dm-VSASL. In vivo experiments using dual-module VSI (dm-VSI) were performed to compare two BS strategies: a conventional one with additional BS pulses and a new one without any BS pulse. Their BS performance, temporal noise, and temporal SNR were examined and compared, with pulsed and pseudo-continuous ASL (PASL and PCASL) as the reference. Results: The in vivo experiments validated the BS modeling. Strong positive linear correlations (r >0.82, p<0.0001) between the temporal noise and the tissue signal were found in PASL/PCASL and dm-VSI. Optimal BS can be achieved with and without additional BS pulses in dm-VSI; the latter improved the ASL signals by 8.5% in gray matter (p=0.006) and 12.2% in white matter (p=0.014) and tended to provide better temporal SNR. The dm-VSI measured significantly higher ASL signal (p<0.016) and temporal SNR (p<0.018) than PASL and PCASL. Complex reconstruction was found necessary with aggressive BS. Conclusion: Guided by modeling, optimal BS can be achieved without any BS pulse in dm-VSASL, further improving the ASL signal and the SNRperformance. [ABSTRACT FROM AUTHOR]

The task of background suppression in infrared small-target scenarios aims to eliminate irregular noisy backgrounds while preserving targets with high-frequency features. In infrared small-target scenes at long distances, the backgrounds become complex and the target features are degraded, highlighting a significant disparity between the detailed and realistic background and the limited features of the targets. To address these challenges, we propose a patch-based semantically enhanced generative adversarial network (GAN) named PSEnet for background suppression in infrared small-target scenarios. First, we introduce a patch-scale GAN that allows the model to concentrate on local background suppression. This shift from a global to local perspective simplifies the complexity of background suppression. Second, we employ the PSE module consisting multiscale dilated convolution and adaptive weight fusion to extract local semantic information. Third, by segmenting the infrared image into smaller patches and resampling them, we create a more balanced dataset for adversarial training. Experimental results demonstrate that the proposed algorithm significantly improves the signal-to-noise ratio of dim and small targets, reduces the missing detection rate, and achieves a precision of almost 91%. In conclusion, this approach effectively uses GANs for background suppression in complex environments.

Hyperspectral target detection (HTD) aims to identifying targets within a hyperspectral image (HSI) based on provided target spectra. In the current HTD field, representation-based detectors have attracted much attention. However, there are two prominent challenges that are particularly noteworthy. First, the background class encompasses diverse land covers, making its accurate representation challenging. Second, the detection ability can be significantly influenced by the abnormalities and noise in HSI. To tackle these concerns, we propose a novel background covariance discriminative dictionary learning (BCDDL) model for HTD. To enhance the background representation ability and overcome the sparse noise, we combine the dictionary learning with spectral covariance descriptors and undertake background reconstruction in regional scale. Specifically, the input HSI is pre-processed into superpixels, the spectral covariance of each superpixel is used to provide a compact and flexible description of local regional statistical properties. Further, a novel spatial clustering-based dictionary learning method is proposed to learn the background discriminative covariances dictionary. The collaborative representation model within symmetric positive definite (SPD) manifold is utilized to reconstruct background region and get the residual. By merging the background residual with pixel-wise target reconstruction residual, we derive final detection output. Comprehensive experiments on two public hyperspectral datasets and two novel GaoFen-5 datasets demonstrate the superiority of our BCDDL approach over 10 state-of-the-art methods, especially in terms of suppressing background.

Vehicle detection based on unmanned aerial vehicle (UAV) aerial images plays a significant role in areas such as traffic monitoring and management, disaster relief, and more, garnering extensive attention from researchers in recent years. However, datasets acquired from UAV platforms inevitably suffer from issues such as imbalanced class distribution, severe background interference, numerous small objects, and significant target scale variance, presenting substantial challenges to practical vehicle detection applications based on this platform. Addressing these challenges, this paper proposes an object detection model grounded in a background suppression pyramid network and multi-scale task adaptive decoupled head. Firstly, the model implements a long-tail feature resampling algorithm (LFRA) to solve the problem of imbalanced class distribution in the dataset. Next, a background suppression pyramid network (BSPN) is integrated into the Neck segment of the model. This network not only reduces the interference of redundant background information but also skillfully extracts features of small target vehicles, enhancing the ability of the model to detect small objects. Lastly, a multi-scale task adaptive decoupled head (MTAD) with varied receptive fields is introduced, enhancing detection accuracy by leveraging multi-scale features and adaptively generating relevant features for classification and detection. Experimental results indicate that the proposed model achieves state-of-the-art performance on lightweight object detection networks. Compared to the baseline model PP-YOLOE-s, our model improves the A P 50 : 95 on the VisDrone-Vehicle dataset by 1.9%. [ABSTRACT FROM AUTHOR]

In this article, we design a new lightweight infrared optical system that fully meets airborne settings and greatly reduces the collection of invalid information. This new system targets the technical problems of stray light, strong invalid information, weak texture information of small targets, and low intensity of valid information under a complex background, which lead to difficult identification of small targets. Image enhancement of weak, small targets against complex backgrounds has been the key to improving small-target search and tracking technology. For the complex information that is still collected, an improved two-channel image enhancement processing algorithm is proposed: the A-channel adopts an improved nonlinear diffusion method and improved curvature filtering, and the B-channel adopts bootstrap filtering and a local contrast enhancement algorithm. The weak target is then extracted by the algorithm of weighted superposition. The false alarm rate is effectively weakened, and robustness is improved. As a result of the experimental data analysis, the method can effectively extract the weak targets in complex backgrounds, such as artificial backgrounds, surface vegetation, etc., enlarge the target gray value, and reduce Fa by 56%, compared with other advanced methods, while increasing Pd by 17%. The algorithm proposed in this paper is of great significance and value for weak target identification and tracking, and it has been successfully applied to industrial detection, medical detection, and in the military field. [ABSTRACT FROM AUTHOR]

Materials used to construct magic‐angle‐spinning NMR probes can contain NMR active nuclei that produce a significant amount of background signal. Because these materials are located outside the sample coil, the use of spatially selective pulses to remove the background is a popular approach for background suppression. However, previously suggested spatially selective pulses suffer from limited excitation bandwidths, which may make them unsuitable for the acquisition of nuclei with a large chemical shift range. Here, a pulse (OC‐BACK) is presented, which has been developed by optimal control, which has a flat profile of ~120 kHz with respect to off‐resonance effects and extended pass and suppression bands with respect to the nominal nutation frequency. The presented solution is large enough to be effective for background suppression schemes in 19F magic‐angle‐spinning NMR at medium and low magnetic fields. [ABSTRACT FROM AUTHOR]

Infrared small-target enhancement in complex contexts is one of the key technologies for infrared search and tracking systems. The effect of enhancement directly determines the reliability of the monitoring equipment. To address the problem of the low signal-to-noise ratio of small infrared moving targets in complex backgrounds and the poor effect of traditional enhancement algorithms, an accurate enhancement method for small infrared moving targets based on two-channel information is proposed. For a single frame, a modified curvature filter is used in the A channel to weaken the background while an improved PM model is used to enhance the target, and a modified band-pass filter is used in the B channel for coarse enhancement followed by a local contrast algorithm for fine enhancement, based on which a weighted superposition algorithm is used to extract a single-frame candidate target. The results of the experimental data analysis prove that the method has a good enhancement effect and robustness for small IR motion target enhancement in complex backgrounds, and it outperforms other advanced algorithms by about 43.7% in ROC. [ABSTRACT FROM AUTHOR]

The monitoring of space debris is important for spacecraft such as satellites operating in orbit, but the background in star images taken by ground-based telescopes is relatively complex, including stray light caused by diffuse reflections from celestial bodies such as the Earth or Moon, interference from clouds in the atmosphere, etc. This has a serious impact on the monitoring of dim and small space debris targets. In order to solve the interference problem posed by a complex background, and improve the signal-to-noise ratio between the target and the background, in this paper, we propose a novel star image enhancement algorithm, MBS-Net, based on background suppression. Specifically, the network contains three parts, namely the background information estimation stage, multi-level U-Net cascade module, and recursive feature fusion stage. In addition, we propose a new multi-scale convolutional block, which can laterally fuse multi-scale perceptual field information, which has fewer parameters and fitting capability compared to ordinary convolution. For training, we combine simulation and real data, and use parameters obtained on the simulation data as pre-training parameters by way of parameter migration. Experiments show that the algorithm proposed in this paper achieves competitive performance in all evaluation metrics on multiple real ground-based datasets. [ABSTRACT FROM AUTHOR]

It is difficult for traditional algorithms to remove cloud edge contours in multi-cloud scenarios. In order to improve the detection ability of dim and small targets in complex edge contour scenes, this paper proposes a new dim and small target detection algorithm based on local multi-directional gradient information energy perception. Herein, based on the information difference between the target area and the background area in the four direction neighborhood blocks, an energy enhancement model for multi-directional gray aggregation (EMDGA) is constructed to preliminarily enhance the target signal. Subsequently, a local multi-directional gradient reciprocal background suppression model (LMDGR) was constructed to model the background of the image. Furthermore, this paper proposes a multi-directional gradient scale segmentation model (MDGSS) to obtain candidate target points and then combines the proposed multi-frame energy-sensing (MFESD) detection algorithm to extract the true targets from sequence images. Finally, in order to better illustrate the effect of the algorithm proposed in this paper in detecting small targets in a cloudy background, four sequence images are selected for detection. The experimental results show that the proposed algorithm can effectively suppress the edge contour of complex clouds compared with the traditional algorithm. When the false alarm rate Pf is 0.005%, the detection rate Pd is greater than 95%. [ABSTRACT FROM AUTHOR]

The accuracy of hyperspectral target detection is often affected by the problems of spectral variation and complex background distribution. Inspired by the powerful representational ability of deep learning, we proposed a three-dimensional (3-D) convolution-based global spatial–spectral attention network (GS2A-Net) to deal with spectral variation in hyperspectral images (HSIs). GS2A-Net uses 3-D convolution kernels of different sizes to capture local spatial and spectral features to achieve multiscale information interaction. Different from the previous 2-D attention mechanisms, GS2A-Net simultaneously considers the information in the spatial and spectral dimensions, and creates a weight map consistent with the size of the original HSI. Furthermore, we proposed a new background suppression strategy based on the spectral angle mapping to achieve more accurate target detection, which can preserve the targets as much as possible when suppressing the background. The method was validated through experiments on five real-world HSI datasets. Compared with several classical and deep-learning-based methods, the proposed method exhibits greater detection accuracy.

Vehicle detection based on unmanned aerial vehicle (UAV) aerial images plays a significant role in areas such as traffic monitoring and management, disaster relief, and more, garnering extensive attention from researchers in recent years. However, datasets acquired from UAV platforms inevitably suffer from issues such as imbalanced class distribution, severe background interference, numerous small objects, and significant target scale variance, presenting substantial challenges to practical vehicle detection applications based on this platform. Addressing these challenges, this paper proposes an object detection model grounded in a background suppression pyramid network and multi-scale task adaptive decoupled head. Firstly, the model implements a long-tail feature resampling algorithm (LFRA) to solve the problem of imbalanced class distribution in the dataset. Next, a background suppression pyramid network (BSPN) is integrated into the Neck segment of the model. This network not only reduces the interference of redundant background information but also skillfully extracts features of small target vehicles, enhancing the ability of the model to detect small objects. Lastly, a multi-scale task adaptive decoupled head (MTAD) with varied receptive fields is introduced, enhancing detection accuracy by leveraging multi-scale features and adaptively generating relevant features for classification and detection. Experimental results indicate that the proposed model achieves state-of-the-art performance on lightweight object detection networks. Compared to the baseline model PP-YOLOE-s, our model improves the AP50:95 on the VisDrone-Vehicle dataset by 1.9%.

Most background suppression algorithms are weakly robust due to the complexity and fluctuation of the star image's background. In this paper, a background suppression algorithm for stray lights in star images is proposed, which is named BSC-Net (Background Suppression Convolutional Network) and consist of two parts: "Background Suppression Part" and "Foreground Retention Part". The former part achieves background suppression by extracting features from various receptive fields, while the latter part achieves foreground retention by merging multi-scale features. Through this two-part design, BSC-Net can compensate for blurring and distortion of the foreground caused by background suppression, which is not achievable in other methods. At the same time, a blended loss function of smooth_L1&Structure Similarity Index Measure (SSIM) is introduced to hasten the network convergence and avoid image distortion. Based on the BSC-Net and the loss function, a dataset consisting of real images will be used for training and testing. Finally, experiments show that BSC-Net achieves the best results and the largest Signal-to-Noise Ratio (SNR) improvement in different backgrounds, which is fast, practical and efficient, and can tackle the shortcomings of existing methods. [ABSTRACT FROM AUTHOR]

At present, poor background suppression is one major problem for visual saliency detection. Although many mainstream saliency detection models can effectively locate salient objects, objects in complicated backgrounds in some natural images are often mistaken for salient objects. Therefore, this paper proposed to set up a center prior-based encoder-decoder network to improve background suppression results. A traditional center prior-based method and U-Net model were combined efficiently. First, multi-scale group convolution was used to replace general convolution, which can highlight the semantic information characteristics, and high-level characteristics at the bottom of U-Net were integrated and optimized on the basis of the consideration of center prior. Then, refinements were delivered throughout the whole network by upgrading the network structure, so as to ensure the optimized features can be made full use of. Since the changes to the U-Net architecture somewhat affected the stability of the network. Therefore, branch network modules were adopted and adaptive parameters defined to coordinate the relationships between branch networks to keep the network structure well balanced. The method has been tested with four widely used databases and proven effective by comparing its results with those of another seven popular methods. [ABSTRACT FROM AUTHOR]

In order to reduce the influence of strong halo effect on dim small target detection in daytime, a dim small target detection algorithm based on spiral gradient optimization estimation and high-order correlation enhancement is proposed in this paper. In this paper, we first design a spiral motion model to obtain the local gradient information in the central image point by perturbing the designed motion direction, then estimate the optimal background by establishing a gradient optimization model to achieve background suppression while effectively removing the halo phenomenon. Considering the original high-order correlation model only uses a single pixel for motion correlation, there is insufficient information utilization, an improved high-order correlation energy enhancement model is proposed to enhance the target signal, the algorithm first constructs an attention discrimination model based on inner and outer windows to obtain the salient region of the image, and then carries out multi frame high-order motion correlation of neighborhood blocks to enhance the target energy. After experiments, it is shown that compared with the traditional algorithm, the algorithm proposed in this paper can effectively weaken the halo effect while suppressing most of the background and can effectively enhance the local signal-to-noise ratio of the target.

Small target detection from the infrared remote sensing image is a challenge task. In this article, a novel local adaptive threshold algorithm combined with heterogeneity and compactness filters is proposed to detect the small target from the infrared remote sensing images. First, the infrared image is filtered by a heterogeneity filter to enhance the target saliency. Then, the enhanced image is filtered by a compactness filter to generate a target candidate region map. Finally, for each pixel in the target candidate region, a local adaptive threshold is calculated from the enhanced image to determine whether it is a target pixel or not, and thus, the targets are extracted out. The designed heterogeneity filter and compactness filter can effectively suppress the background clutter, enhance the target, and generate target candidate regions. The proposed adaptive thresholding is a local threshold method, which is calculated in a small local window and can effectively reduce the false alarm and missing alarm. Qualitative and quantitative experiments are conducted on synthetic images and real images. The experiment results show that, with good target enhancement and background suppression, and high detection accuracy, the proposed method outperforms other state-of-the-art methods.

Infrared small-target enhancement in complex contexts is one of the key technologies for infrared search and tracking systems. The effect of enhancement directly determines the reliability of the monitoring equipment. To address the problem of the low signal-to-noise ratio of small infrared moving targets in complex backgrounds and the poor effect of traditional enhancement algorithms, an accurate enhancement method for small infrared moving targets based on two-channel information is proposed. For a single frame, a modified curvature filter is used in the A channel to weaken the background while an improved PM model is used to enhance the target, and a modified band-pass filter is used in the B channel for coarse enhancement followed by a local contrast algorithm for fine enhancement, based on which a weighted superposition algorithm is used to extract a single-frame candidate target. The results of the experimental data analysis prove that the method has a good enhancement effect and robustness for small IR motion target enhancement in complex backgrounds, and it outperforms other advanced algorithms by about 43.7% in ROC.

Background suppression (BGS) in arterial spin labeling (ASL) magnetic resonance imaging leads to a higher temporal signal‐to‐noise ratio (tSNR) of the perfusion images compared with ASL without BGS. The performance of the BGS, however, depends on the tissue relaxation times and on inhomogeneities of the scanner's magnetic fields, which differ between subjects and are unknown at the moment of scanning. Therefore, we developed a feedback loop (FBL) mechanism that optimizes the BGS for each subject in the scanner during acquisition. We implemented the FBL for 2D pseudo‐continuous ASL scans with an echo‐planar imaging readout. After each dynamic scan, the acquired ASL images were automatically sent to an external computer and processed with a Python processing tool. Inversion times were optimized on the fly using 80 iterations of the Nelder–Mead method, by minimizing the signal intensity in the label image while maximizing the signal intensity in the perfusion image. The performance of this method was first tested in a four‐component phantom. The regularization parameter was then tuned in six healthy subjects (three males, three females, age 24–62 years) and set as λ = 4 for all other experiments. The resulting ASL images, perfusion images, and tSNR maps obtained from the last 20 iterations of the FBL scan were compared with those obtained without BGS and with standard BGS in 12 healthy volunteers (five males, seven females, age 24–62 years) (including the six volunteers used for tuning of λ). The FBL resulted in perfusion images with a statistically significantly higher tSNR (2.20) compared with standard BGS (1.96) (p<5x10−3, two‐sided paired t‐test). Minimizing signal in the label image furthermore resulted in control images, from which approximate changes in perfusion signal can directly be appreciated. This could be relevant to ASL applications that require a high temporal resolution. Future work is needed to minimize the number of initial acquisitions during which the performance of BGS is reduced compared with standard BGS, and to extend the technique to 3D ASL. [ABSTRACT FROM AUTHOR]

It is difficult for traditional algorithms to remove cloud edge contours in multi-cloud scenarios. In order to improve the detection ability of dim and small targets in complex edge contour scenes, this paper proposes a new dim and small target detection algorithm based on local multi-directional gradient information energy perception. Herein, based on the information difference between the target area and the background area in the four direction neighborhood blocks, an energy enhancement model for multi-directional gray aggregation (EMDGA) is constructed to preliminarily enhance the target signal. Subsequently, a local multi-directional gradient reciprocal background suppression model (LMDGR) was constructed to model the background of the image. Furthermore, this paper proposes a multi-directional gradient scale segmentation model (MDGSS) to obtain candidate target points and then combines the proposed multi-frame energy-sensing (MFESD) detection algorithm to extract the true targets from sequence images. Finally, in order to better illustrate the effect of the algorithm proposed in this paper in detecting small targets in a cloudy background, four sequence images are selected for detection. The experimental results show that the proposed algorithm can effectively suppress the edge contour of complex clouds compared with the traditional algorithm. When the false alarm rate Pf is 0.005%, the detection rate Pd is greater than 95%.

The monitoring of space debris is important for spacecraft such as satellites operating in orbit, but the background in star images taken by ground-based telescopes is relatively complex, including stray light caused by diffuse reflections from celestial bodies such as the Earth or Moon, interference from clouds in the atmosphere, etc. This has a serious impact on the monitoring of dim and small space debris targets. In order to solve the interference problem posed by a complex background, and improve the signal-to-noise ratio between the target and the background, in this paper, we propose a novel star image enhancement algorithm, MBS-Net, based on background suppression. Specifically, the network contains three parts, namely the background information estimation stage, multi-level U-Net cascade module, and recursive feature fusion stage. In addition, we propose a new multi-scale convolutional block, which can laterally fuse multi-scale perceptual field information, which has fewer parameters and fitting capability compared to ordinary convolution. For training, we combine simulation and real data, and use parameters obtained on the simulation data as pre-training parameters by way of parameter migration. Experiments show that the algorithm proposed in this paper achieves competitive performance in all evaluation metrics on multiple real ground-based datasets.

A new algorithm for hyperspectral image anomaly detection is proposed by designing an adaptive multi-layer structure with spatial-spectral combination information, which is different from the traditional anomaly detection algorithms only considering the spectral difference between the anomaly point and the background pixels, and ignoring the difference between the local spatial structure and spectrum. Firstly, the present algorithm not only calculates the spectral dimension difference between the pixels to be measured and the pixels in the background window, but also measures the spatial structure difference between the internal window and the background window. Mostly, an adaptive multi-layer structure for anomaly detection framework is carried out based on the idea of background suppression, and a multi-layered anomaly detector is constructed. The anomaly detection results of each layer of the detector are taken as the constraints, and the background information of the image input in the detector of the next layer is suppressed, adaptively suppressing the background noises. The experimental results show that the present algorithm makes better use of both the local spatial structure and the spectral dimension information than the traditional two-window models (global RX, local RX and KRX), adaptively suppresses background, reduces the false alarm rate, and improves the detection effect of the abnormal targets with fewer pixels.

The offline Siamese tracking methods have recently attracted much attention due to their impressive performance on speed and accuracy. However, they ignore the target and background information during tracking, thus lacking sufficient discrimination ability to handle complex background interference, and cannot robustly adapt to sharp appearance variations in challenging situations, such as scale variation, rotation, occlusion, and background clutter. To address the drawbacks above, we propose an online discrimination-correction subnet, which discriminates and corrects the tracking results obtained by the Siamese tracker through online learning, and further guides the template update. Firstly, a feature refinement module is designed to enhance the discriminative ability of features, which compresses and combines target features based on the importance of feature channels to obtain target-specific features. On this basis, an online discriminator module is proposed to learn the decision boundary between target and background through samples collected online to suppress background interference. In addition, a sample equilibrium strategy is proposed to alleviate the imbalance in the distribution of positive and negative samples during online training. Finally, to handle sharp target appearance variations, a robust template updating strategy is proposed to screen high-quality historical samples to generate dynamic templates. To evaluate the generalization ability and effectiveness of our method, we apply it to four classical Siamese trackers, namely SiamFC, SiamRPN++, SiamCAR, and TransT. Extensive experiments on seven challenging benchmarks demonstrate that our method significantly improves the baseline methods and achieves outstanding performance compared with state-of-the-art trackers. • A feature refining module for obtaining target-special features. • Online discriminative learning for background suppression. • Sample equilibrium strategy for alleviating the imbalance of online samples. • Dynamic templates for adapting to target appearance variations. [ABSTRACT FROM AUTHOR]

• Introduces polarization modulation in SIM to suppress background. • Enhances structural fidelity and resolution through multidirectional fusion. • Addresses key SIM limitations, improving signal localization and image quality. We report a structured illumination microscopy (SIM) reconstruction method based on polarization modulation (PM), which can suppress background noise by utilizing the polarization-sensitivity properties of fluorescent dipoles. The proposed method employs a ferroelectric liquid crystal (FLC) spatial light modulator (SLM) to generate multidirectional interference patterns, and uses a combination of two liquid crystal variable phase retarders (LCVRs) and a quarter-wave plate (QWP) for flexible and rapid polarization modulation. The results of simulations and experiments demonstrate that the combination of SIM and PM effectively suppresses background noise in SIM with a slight improvement in resolution. This technique also solves the problem of structural distortion that occasionally exists in SIM and improves the accuracy of the reconstruction. [ABSTRACT FROM AUTHOR]

• Proposed infrared differential detection (IDD) method obtains multiple-channel characteristic differences. • Proposed band selection method for IDD can significantly improve space-based aerial target detection performance. • A better detection performance can be obtained in IDD despite using a band with low SCR. • Background characteristics have an important influence on the IDD band selection. Space-based aerial target detection is an important branch of infrared remote sensing. Due to the complex background and dim target, the problem of low target detectability needs to be solved. This work introduces a novel infrared differential detection (IDD) method, which acquires information differences under multiple channels in a dimension. The paper further proposes a band selection methodology based on spectral radiation characteristics of the target, background, and clutter. Firstly, the infrared differential analytic factor (IDAF) model is constructed to represent the target-background contrast. Secondly, the dual-band signal-to-clutter ratio (DBSCR) is proposed to evaluate the target observability. Next, the IDAF-DBSCR objective function selects the optimal dual bands. The experiment sets up two typical aerial targets and eight detection scenes for verification, and the selected bands can effectively suppress the background and be applied to general scenarios. The simulation results show an order-of-magnitude improvement in SCR quantified by the gain of the signal-to-clutter ratio (GSCR), and the performance of dual-band detection is significantly better than that of single-band detection. [ABSTRACT FROM AUTHOR]

To address the issues of signal saturation and measurement range reduction caused by the background in light-induced thermoelastic spectroscopy (LITES) employing the first harmonic (1 f) modulation, an optical-domain modulation cancellation method (MOCAM) was reported in the 1 f -based LITES. A balance laser source with the opposite sine modulation phase as the excitation laser source synchronously excites the quartz tuning fork (QTF) to realize background suppression. By optimizing the modulation amplitude of the balance laser and the modulation phase difference between the two laser sources, the background level was suppressed by two orders of magnitude from ∼ 7 mV to ∼ 18 μV, close to the QTF thermal noise level (∼ 1 μV). Combining the excitation laser with the balance laser enables MOCAM to overcome common mode interferences, e.g. optical path shift, laser power fluctuation, and focus position change. Furthermore, methane and water vapor (CH 4 /H 2 O) dual-gas detection based on MOCAM and time-division multiplexing method was demonstrated, and measurement range is enlarged with easy calibration due to zero background. The MOCAM-based 1 f -LITES technique delivers effective background suppression and provides robustness to common mode interferences, thus establishes a feasible solution for mobile vehicle or airborne gas monitoring applications with serious environmental disturbances. • A 1 f -based MOCAM-LITES was proposed as a novel background-suppression technique. • The balance requirements in MOCAM-LITES were derived. • MOCAM's effectiveness is unaffected by common mode factors. • MOCAM decreases the signal saturation likelihood and expands measurement range. • The first dual-gas detection based on MOCAM was reported. [ABSTRACT FROM AUTHOR]

Weakly-supervised temporal action localization aims to identify all action instances and their corresponding categories in the untrimmed videos. Since it involves only videolevel labels during training, resulting in this problem being more challenging. Existing attention-based action localization methods use the attention module to identify action segments and assign them to the appropriate action categories. However, such methods inevitably suffer from many background segments that are similar to the target actions, being recognized as actions. To address this issue, we propose a new weakly-supervised temporal action localization network using background suppression (BS-WTAL). The network defines the filtering module, which can suppress the activation of the background regions, classification module, which identifies the activity categories, and generative attention module, which is learned to model a segment-wise representation. This enables BS-WTAL to accurately distinguish actions from the background. Furthermore, we conduct ablation studies from different perspectives. Extensive experiments are performed on two datasets: THUMOS14 and ActivityNet1.2. Our approach exhibits better performance on these two datasets and achieves performance comparable to the state-of-the-art fully-supervised methods. [ABSTRACT FROM AUTHOR]

To improve the detection of dim and small targets in infrared (IR) images containing high-intensity cloud clutter, a novel adaptive background suppression method is proposed. By using three-dimensional cooperative filtering and differential calculation, the different and complex background clutter is suppressed. To obtain the optimal parameters for the background suppression algorithm, an adaptive parameter optimization method is proposed. The adaptive parameter optimization problem is transformed into a multiobjective optimization problem in which the signal-to-clutter ratio gain and background suppression factor, which effectively reflect the background suppression performance, are chosen as the optimization objectives, and the parameters of the proposed background suppression algorithm are considered as the variables. To effectively solve the established multiobjective optimization problem, a particle swarm parameter optimization-based method is utilized. Experimental results indicate that the proposed adaptive background suppression method using these optimal parameters has good performance for IR images in real complex scenes, as well as performance superior to that of other baseline methods.

Purpose: Background suppression (BS) is recommended in arterial spin labeling (ASL) for improved SNR but is difficult to optimize in existing velocity-selective ASL (VSASL) methods. Dual-module VSASL (dm-VSASL) enables delay-insensitive, robust, and SNR-efficient perfusion imaging, while allowing efficient BS, but its optimization has yet to be thoroughly investigated., Methods: The inversion effects of the velocity-selective labeling pulses, such as velocity-selective inversion (VSI), can be used for BS, and were modeled for optimizing BS in dm-VSASL. In vivo experiments using dual-module VSI (dm-VSI) were performed to compare two BS strategies: a conventional one with additional BS pulses and a new one without any BS pulse. Their BS performance, temporal noise, and temporal SNR were examined and compared, with pulsed and pseudo-continuous ASL (PASL and PCASL) as the reference., Results: The in vivo experiments validated the BS modeling. Strong positive linear correlations (r > 0.82, p < 0.0001) between the temporal noise and the tissue signal were found in PASL/PCASL and dm-VSI. Optimal BS can be achieved with and without additional BS pulses in dm-VSI; the latter improved the ASL signals by 8.5% in gray matter (p = 0.006) and 12.2% in white matter (p = 0.014) and tended to provide better temporal SNR. The dm-VSI measured significantly higher ASL signal (p < 0.016) and temporal SNR (p < 0.018) than PASL and PCASL. Complex reconstruction was found necessary with aggressive BS., Conclusion: Guided by modeling, optimal BS can be achieved without any BS pulse in dm-VSASL, further improving the ASL signal and the SNR performance., (© 2024 The Authors. Magnetic Resonance in Medicine published by Wiley Periodicals LLC on behalf of International Society for Magnetic Resonance in Medicine.)

The widespread screen sharing poses a great challenge in the visual quality control of industrial applications, as there is a significant difference between the visual quality perception of natural content (NC) and screen content (SC) images (or videos) during acquisition, compression, distribution, and display. Apart from the foreground distortion, artifacts appeared in the smooth background of SC images also play a key role in its perceived quality. In this article, a robust approach is designed for quality measurement of SC images, which embodies both foreground perception and background suppression (FPBS) in a satisfying way. Specifically, the extraction of foreground perceptual features is investigated by considering both the local contrast variation and the local edge response features, while for the first time, a suppression strategy is devised to eliminate the influences of artifacts appeared in the simple and homogeneous background. Experimental results show that, based on only two simple features, the proposed scheme achieves the state-of-the-art results on two latest benchmark databases. We believe that the feature engineering method can further improve the performance of the FPBS framework on the quality measurement of screen images. [ABSTRACT FROM AUTHOR]

Unsupervised domain adaptation has been a popular approach for cross-domain person re-identification (re-ID). There are two solutions based on this approach. One solution is to build a model for data transformation across two different domains. Thus, the data in source domain can be transferred to target domain where re-ID model can be trained by rich source domain data. The other solution is to use target domain data plus corresponding virtual labels to train a re-ID model. Constrains in both solutions are very clear. The first solution heavily relies on the quality of data transformation model. Moreover, the final re-ID model is trained by source domain data but lacks knowledge of the target domain. The second solution in fact mixes target domain data with virtual labels and source domain data with true annotation information. But such a simple mixture does not well consider the raw information gap between data of two domains. This gap can be largely contributed by the background differences between domains. In this paper, a Suppression of Background Shift Generative Adversarial Network (SBSGAN) is proposed to mitigate the gaps of data between two domains. In order to tackle the constraints in the first solution mentioned above, this paper proposes a Densely Associated 2-Stream (DA-2S) network with an update strategy to best learn discriminative ID features from generated data that consider both human body information and also certain useful ID-related cues in the environment. The built re-ID model is further updated using target domain data with corresponding virtual labels. Extensive evaluations on three large benchmark datasets show the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Purpose: To develop an improved reconstruction method, k-space subtraction with phase and intensity correction (KSPIC), for highly accelerated, subtractive, non-contrast-enhanced MRA. Methods: The KSPIC method is based on k-space subtraction of complex raw data. It applies a phase-correction procedure to restore the polarity of negative signals caused by subtraction and an intensity-correction procedure to improve background suppression and thereby sparsity. Ten retrospectively undersampled data sets and 10 groups of prospectively undersampled data sets were acquired in 12 healthy volunteers. The performance of KSPIC was compared with another improved reconstruction based on combined magnitude subtraction, as well as with conventional k-space subtraction reconstruction and magnitude subtraction reconstruction, both using quantitative metrics and using subjective quality scoring. Results: In the quantitative evaluation, KSPIC had the best performance in terms of peak SNR, structural similarity index measure, contrast-to-noise ratio of artery-to-background and sharpness, especially at high acceleration factors. The KSPIC method also had the highest subjective scores for all acceleration factors in terms of vessel delineation, image noise and artifact, and background contamination. The acquisition can be accelerated by a factor of 20 without significant decreases of subjective scores. The optimal size of the phase-correction region was found to be 12-20 pixels in this study. Conclusion: Compared with combined magnitude subtraction and conventional reconstructions, KSPIC has the best performance in all of the quantitative and qualitative measurements, permitting good image quality to be maintained up to higher accelerations. The KSPIC method has the potential to further reduce the acquisition time of subtractive MRA for clinical examinations. [ABSTRACT FROM AUTHOR]

Most background suppression algorithms are weakly robust due to the complexity and fluctuation of the star image’s background. In this paper, a background suppression algorithm for stray lights in star images is proposed, which is named BSC-Net (Background Suppression Convolutional Network) and consist of two parts: “Background Suppression Part” and “Foreground Retention Part”. The former part achieves background suppression by extracting features from various receptive fields, while the latter part achieves foreground retention by merging multi-scale features. Through this two-part design, BSC-Net can compensate for blurring and distortion of the foreground caused by background suppression, which is not achievable in other methods. At the same time, a blended loss function of smooth_L1&Structure Similarity Index Measure (SSIM) is introduced to hasten the network convergence and avoid image distortion. Based on the BSC-Net and the loss function, a dataset consisting of real images will be used for training and testing. Finally, experiments show that BSC-Net achieves the best results and the largest Signal-to-Noise Ratio (SNR) improvement in different backgrounds, which is fast, practical and efficient, and can tackle the shortcomings of existing methods.

The double beta decay of 136Xe to excited states of 136Ba (DBD-ES) has not yet been discovered experimentally. The experimental signature of such decays, one or two gamma rays following the beta signals, can be identified more effectively in a gaseous detector with the help of topological signatures. We have investigated key parameters of particle trajectories of DBD-ES with Monte Carlo simulation data of the proposed PandaX-III detector as an example. The background rates can be reduced by about one order of magnitude while keeping more than half of signals with topological analysis. The estimated half-life sensitivity of DBD-ES can be improved by 1.8 times to 4.1×1023 year (90% C.L.). Similarly, the half-life sensitivity of neutrinoless double beta decay of 136Xe to excited states of 136Ba can be improved by a factor of 4.8 with topological signatures. [ABSTRACT FROM AUTHOR]

In recent years, deep convolutional features have been applied to discriminative correlation filters-based methods, which have achieved impressive performance in tracking. Most of them utilize hierarchical features from a certain layer. However, this is not always sufficient to learn target appearance changes and to suppress the background interference in complicated interfering factors (e.g., deformation, fast motion, low resolution, and rotations). In this paper, we propose an adaptive multi-branch correlation filter tracking method, by constructing multi-branch models and using an adaptive selection strategy to improve the accuracy and robustness of visual tracking. Specially, the multi-branch models are introduced to tolerate temporal changes of the object, which can serve different circumstances. In addition, the adaptive selection strategy incorporates both foreground and background information to learn background suppression. To further improve the tracking performance, we propose a measurement method to handle tracking failures from unreliable samples. Extensive experiments on OTB-2013, OTB-2015, and VOT-2016 datasets show that the proposed tracker has comparable performance compared to state-of-the-art tracking methods. Especially, on the OTB-2015, our method significantly improves the baseline with a gain of 5.5% in overlap precision. [ABSTRACT FROM AUTHOR]

Although significant progress has been made in saliency detection, predicting saliency remains challenging when the scene is complex, especially when salient and non-salient regions are similar or salient objects have intricate contours. Previous advanced methods rarely explored learning in the background of images. In fact, background and foreground of an image contain complementary information. In this work, we propose to decompose the saliency detection task into three subtasks: foreground awareness, background suppression, and edge refinement. More specifically, our decoder is comprised of three branches: a foreground awareness branch, a background suppression branch, and an edge refinement branch. Each branch aims to learn specific features for predicting its corresponding map. Meanwhile, we design a regional focus loss function with controllable modulating factors to supervise the learning of each branch at the training stage. Moreover, we build an attention guided feature fusion module to adaptively fuse multi-scale features and a global information capture module to locate salient objects. Experiments on five benchmark datasets demonstrate that our approach is superior to the state-of-the-art methods. • We divide salient object detection into three subtasks to focus on different regions. • We design a regional focus loss to supervise subtask learning. • We propose a global information capture module for precise object localization. • We design an attention guided feature fusion module to fuse multi-scale features. [ABSTRACT FROM AUTHOR]

• Fluorescence Diagnosis with violet excitation • Contrast enhancement methods • Contrast can be increased by Red/Green and Red–Green methods • Ambient light suppression by time-resolved subtraction from the fluorescence frame • Endogenous fluorescence is suppressed by contrast enhancement methods Two Bispectral contrast enhancement approaches for the fluorescence diagnosis with chlorine-e6 and a wide field-of-view imaging system with fluorescence excitation at 405 nm and time-resolved background suppression were analyzed and compared. Two techniques for the contrast enhancement of a fluorescent video system (Red/Green (R/G) ratio and Red-Green (R-G)) with time-resolved background suppression for fluorescent diagnosis (FD) were tested in four patients with basal cell carcinoma (BCC). The results of both contrast enhancement methods were compared for the diagnostic efficiency for FD of BCC. Both techniques successfully determined the boundaries of the lesions and the fluorescence intensity. Both contrast enhancement modes have proven effective in identifying tumor borders in cases of low contrast in BCC FD with Ce6. While the Red/Green (R/G) mode provides sharper lesion borders, the Red minus Green (R-G) mode visualizes more fluorescent features and makes it easier to assess the lesion margins. [ABSTRACT FROM AUTHOR]

Aiming to recognize persons of interest cross cameras in different locations, the technique of person re-identification (re-ID) has attracted unprecedented attention in the field of public security. However, most of the existing work ignores the influence of background noise and pedestrian's partial information on recognition accuracy. Moreover, the tracking procedure which has a great importance on the real world is often stripped out of the re-ID framework. Therefore, this paper proposes an end-to-end noise-weakened person re-ID and tracking model with adaptive partial information. First, to suppress the background noise and improve the feature discriminability, Mask R-CNN is applied to extract the foreground “pedestrians” out of the complicated background for feature supplement. Second, an adaptive pose estimation model is proposed to make an in-depth analysis of every human body part, thus boosting the robustness against the posture change and individual difference. Finally, to fuse the tracking procedure, a scope prediction scheme based on the pedestrian's moving speed is presented to replace the traditional full domain estimation approach, thus greatly reducing the computational complexity. The extensive experiments have been conducted and the results demonstrate that our method achieves 89.78% and 81.87% rank 1 accuracy on Market-1501 and DukeMTMC-reID with real-time tracking capability, which exhibits great superiority than the state-of-the-art methods.

Purpose: To correct the intensity difference of static background signal between bright blood images and dark blood images in subtractive non‐contrast–enhanced MR angiography using robust regression, thereby improving static background signal suppression on subtracted angiograms. Methods: Robust regression (RR), using iteratively reweighted least squares, is used to calculate the regression coefficient of background tissues from a scatter plot showing the voxel intensity of bright blood images versus dark blood images. The weighting function is based on either the Euclidean distance from the estimated regression line or the deviation angle. Results from RR using the deviation angle (RRDA), conventional RR using the Euclidean distance, and ordinary leastsquares regression were compared with reference values determined manually by two observers. Performance was evaluated over studies using different sequences, including 36 thoracic flow‐sensitive dephasing data sets, 13 iliac flow‐sensitive dephasing data sets, and 26 femoral fresh blood imaging data sets. Results: RR deviation angle achieved robust and accurate performance in all types of images, with small bias, small mean absolute error, and high‐correlation coefficients with reference values. Background tissues, such as muscle, veins, and bladder, were suppressed while the vascular signal was preserved. Euclidean distance gave good performance for thoracic and iliac flow‐sensitive dephasing, but could not suppress background tissues in femoral fresh blood imaging. Ordinary least squares regression was sensitive to outliers and overestimated regression coefficients in thoracic flow‐sensitive dephasing. Conclusion: Weighted subtraction using RR was able to acquire the regression coefficients of background signal and improve background suppression of subtractive non‐contrast–enhanced MR angiography techniques. RR deviation angle has the most robust and accurate overall performance among three regression methods. [ABSTRACT FROM AUTHOR]