Modeling complex rigid motion across large spatiotemporal spans remains an unresolved challenge in dynamic reconstruction. Existing paradigms are mainly confined to short-term, small-scale deformation and offer limited consideration for physical consistency. This study proposes PMGS, focusing on reconstructing Projectile Motion via 3D Gaussian Splatting. The workflow comprises two stages: 1) Target Modeling: achieving object-centralized reconstruction through dynamic scene decomposition and an improved point density control; 2) Motion Recovery: restoring full motion sequences by learning per-frame SE(3) poses. We introduce an acceleration consistency constraint to bridge Newtonian mechanics and pose estimation, and design a dynamic simulated annealing strategy that adaptively schedules learning rates based on motion states. Futhermore, we devise a Kalman fusion scheme to optimize error accumulation from multi-source observations to mitigate disturbances. Experiments show PMGS’s superior performance in reconstructing high-speed nonlinear rigid motion compared to mainstream dynamic methods.

Recently, deep learning network is introduced to terrain segmentation application in polarimetric synthetic aperture radar (PolSAR) images and achieves remarkable performance. However, interclass and intraclass variety caused by nonlinear characteristic and intrinsic randomness introduced by speckle is still a challenge for segmentation methods. In this article, a segmentation framework based on nonlinear statistical learning is introduced to bridge the gap between terrain segmentation in PolSAR and optical images. Firstly, a learnable nonlinear statistical description module is proposed to represent the nonlinear characteristic introduced by coherent speckle, which supplements the lack of discriminative linear feature in terrain blocks. Secondly, a conditional diffusion pipeline is introduced to model latent distributions and perceive intrinsic randomness that causes intraclass variety. Finally, the proposed framework adopts the statistical features learned from nonlinear statistical description module as condition to align with the process of distribution modeling in the conditional diffusion segmentation pipeline. Extensive experiments are conducted on two authoritative PolSAR terrain segmentation datasets, which present competitive performance.

By leveraging the blur-noise trade-off, imaging with non-uniform exposures largely extends the image acquisition flexibility in harsh environments. However, the limitation of conventional cameras in perceiving intra-frame dynamic information prevents existing methods from being implemented in the real-world frame acquisition for real-time adaptive camera shutter control. To address this challenge, we propose a novel Neuromorphic Shutter Control (NSC) system to avoid motion blur and alleviate instant noise, where the extremely low latency of events is leveraged to monitor the real-time motion and facilitate the scene-adaptive exposure. Furthermore, to stabilize the inconsistent Signal-to-Noise Ratio (SNR) caused by the non-uniform exposure times, we propose an event-based image denoising network within a self-supervised learning paradigm, i.e., SEID, exploring the statistics of image noise and inter-frame motion information of events to obtain artificial supervision signals for high-quality imaging in real-world scenes. To illustrate the effectiveness of the proposed NSC, we implement it in hardware by building a hybrid-camera imaging prototype system, with which we collect a real-world dataset containing well-synchronized frames and events in diverse scenarios with different target scenes and motion patterns. Experiments on the synthetic and real-world datasets demonstrate the superiority of our method over state-of-the-art approaches.

Due to the extremely low latency, events have recently been utilized to complement lost information in motion deblurring. Existing approaches largely rely on the perfect pixel-wise alignment between intensity images and events, which usually conflicts with the real world. To tackle this problem, we propose a novel coarse-to-fine framework, named network of event-based motion deblurring with stereo event and intensity cameras (St-EDNet), to recover high-quality images directly from the misaligned inputs that contain both blurry images and the concurrent event stream. Specifically, the coarse spatial alignment of the blurry image and the event stream is first implemented with a cross-modal stereo-matching module without the need for ground-truth depths. Then, a dual-feature embedding architecture is proposed to gradually build the fine bidirectional association of the coarsely aligned data and reconstruct the sequence of the latent sharp images. Furthermore, we build a new dataset with stereo event and intensity cameras (StEIC), containing real-world events, intensity images, and dense disparity maps. Experiments on real-world datasets demonstrate the superiority of the proposed network over state-of-the-art methods. The code and dataset are available at https://mingyuan-lin.github.io/St-ED_web/.

Tracking multiple objects in videos captured by unmanned aerial vehicles is challenging due to sudden viewpoint changes, non-linear motion trajectories, and rapid variations in target size and appearance. Existing methods often struggle to handle these complexities, as they rely heavily on handcrafted geometric constraints and fail to adapt to significant field-of-view changes. To address these issues, this paper presents the Dynamic Field-Aware Multi-Object Tracker (DFA-MOT), a joint detection and tracking framework that integrates detection and motion prediction into a unified model, enhancing tracking performance in dynamic UAV environments. The proposed Dynamic Field-of-View Consistency Learning (DFCL) module mitigates geometric distortions caused by UAV movement by leveraging optical flow and learnable deformation operations to achieve progressive spatial alignment. A Scale-Aware Tracking (SAT) mechanism is explored, which enables to accurately predict of both position and scale variations, enhancing the model’s adaptability to variations in target size. By combining detection with predictive motion modeling, DFA-MOT effectively overcomes the limitations of traditional manual constraints. Extensive experiments on the VisDrone2019 and UAVDT datasets demonstrate that DFA-MOT significantly outperforms state-of-the-art tracking methods in UAV scenarios.

Sleep posture identification is crucial for accurately assessing sleep quality and diagnosing related diseases. In the realm of non-intrusive sleep monitoring, non-contact technologies are becoming increasingly mainstream. Millimeter-wave radar is frequently utilized in sleep posture identification due to its high resolution, strong penetration, and excellent sensitivity. However, traditional radar-based methods for sleep posture identification often struggle with reliability when dealing with diverse individuals and complex sleep environments. To address these challenges, we propose UniSleepPos, which designs a novel dual-view fusion mechanism to integrate depression and elevation angle signals obtained from radar, thus accurately capturing the posture information of the monitored subject in three-dimensional space. Furthermore, we combine sleep posture identification with individual characteristics, utilizing existing individual labels as prior knowledge to assist in sleep posture identification. The integration of prior knowledge provides a valuable information source for the model, helping to enhance its understanding of the data and improve its performance. We collected sleep posture data from eight volunteers using millimeter-wave radar devices under various environmental conditions. Leave-one-subject-out experiments were conducted to validate the effectiveness of UniSleepPos. The results indicated that UniSleepPos significantly outperforms existing methods, demonstrating its potential for practical applications.

Even though the collaboration between traditional and neuromorphic event cameras brings prosperity to frame-event based vision applications, the performance is still confined by the resolution gap crossing two modalities in both spatial and temporal domains. This paper is devoted to bridging the gap by increasing the temporal resolution for images, i.e., motion deblurring, and the spatial resolution for events, i.e., event super-resolving, respectively. To this end, we introduce CrossZoom, a novel unified neural Network (CZ-Net) to jointly recover sharp latent sequences within the exposure period of a blurry input and the corresponding High-Resolution (HR) events. Specifically, we present a multi-scale blur-event fusion architecture that leverages the scale-variant properties and effectively fuses cross-modal information to achieve cross-enhancement. Attention-based adaptive enhancement and cross-interaction prediction modules are devised to alleviate the distortions inherent in Low-Resolution (LR) events and enhance the final results through the prior blur-event complementary information. Furthermore, we propose a new dataset containing HR sharp-blurry images and the corresponding HR-LR event streams to facilitate future research. Extensive qualitative and quantitative experiments on synthetic and real-world datasets demonstrate the effectiveness and robustness of the proposed method.

Vision Transformer has achieved remarkable success in image recognition tasks owing to its global modeling ability. However, the quadratic computational complexity becomes a prominent issue when dealing with high-resolution remote sensing images. Numerous studies have explored the potential of spectral analysis to reveal for more discriminative features. However, neural network exhibit frequency tendency, and different features are interested in different frequencies. Unfortunately, there is no well-established criterion for selecting appropriate frequency representations. To address these issues, a novel wavelet tree head attention (WTHA-ViT) model is proposed which combines a tree structure on the wavelet frequencies with multihead attention in the Transformer encoder, possessing the ability to interact with cross-combinations of short and long-range as well as high and low-frequency components. First, we construct a wavelet tree reduction module (WTRM) based on the wavelet tree structure, utilizing the wavelet decomposition to retain frequency features suitable for each patch, which enables global modeling with various frequency components while reducing computational complexity. Second, guided by channel correlations, we propose the channel lifting scheme multihead attention (CLSMHA) to model the importance on the heads of multihead attention and focus on the more salient head features. Finally, our WTHA-ViT can replace the backbone of detection networks for dense prediction tasks. Extensive experiments on DOTA-V1.0 and HRSID datasets demonstrate that our model exhibits superior performance and robustness compared to state-of-the-art networks. Besides, we evaluate the transferability of the model on DIOR and LEVIR datasets and verify its generalization ability. The code is available athttps://github.com/conquer-pan/WTHA-ViT.

The vanilla tensor in convolutional neural networks (CNNs) can be seen as a mixture of feature information at different frequencies, which is currently only used as a carrier of information. However, few people notice that vanilla tensor is spatial information redundant and information interaction of different frequency bands is beneficial for CNNs. In this paper, we design a novel Wavelet-based frequency-dividing interactive block (WFDI) to factorize a vanilla tensor into a pair of tensors with complementary information to reduce redundancy. Based on this, we embed it into the CNN (WFDI-CNN) for image classification. Specifically, the WFDI-CNN factorizes the vanilla tensor into a low-frequency tensor with lower spatial resolution and a high-frequency tensor with complementary information. Then, the information interaction and forward propagation between the high-frequency and low-frequency tensors not only save computational resources but also improve the network performance. Experimental results on CIFAR10 and CIFAR100 datasets all demonstrate the effectiveness of the proposed WFDI block.

Deep learning advances have propelled common object detection development. However, small object detection in aerial images remains inaccurate. Intersection-over-Union (IoU) has limitations in handling the separation or inclusion cases of prediction and ground truth boxes, especially in small object detection, which significantly hampers the process of bounding box regression. To tackle the issue, we propose Balanced Corner-IoU (BC-IoU) loss, which incorporates both corner point distances and IoU metric, while simultaneously introducing the instance area as a component of loss terms. Moreover, Point Offset Module (POM) branch is developed to generate additional positive samples for small object regression by dynamically controlling anchor point generation. With the above designs, Scale Adaptive Network (SAN) provides a solution to bounding box regression of small objects. Experiments on the small object detection dataset show that BC-IoU loss outperforms other IoU loss variants and that SAN significantly improves performance over the Fully Convolutional One-Stage (FCOS) baseline.

Oriented object detection based on deep learning has received extensive attention while marking the Oriented Bounding Box (OBB) is time-consuming and laborious. For the oriented object detection task, this paper proposes a point-based weakly semi-supervised training strategy, which only requires the training set with an extremely small number (10%) of fully labeled images with OBB and the other with points. Specifically, following the common self-training pipeline, we propose Point to OBB Network (P2ONet) as the teacher model to generate the high-quality pseudo OBB for each point-annotated object. Inspired by channel attention, we introduce Group Attention to P2ONet to better tackle the opposite assignment for similar proposals in the training task leading by the point label assignment strategy. Furthermore, by exploring the constraint in normal self-training pipeline, we propose Confidence-Aware Loss to alleviate the impact of inaccurate pseudo-boxes. Experiments on the DOTA dataset show the close performance between our method and normal oriented object detection training methods with remarkably lower labeling costs.

Homography estimation is an important image alignment method that has been widely used in computer vision applications. Traditional methods heavily rely on the distribution of features and usually fail in low-texture and large-baseline scenes. Most learning-based methods use convolutional neural networks(CNNs) to extract features. However, the dense features extracted in this way have a limited receptive field, leading to poor accuracy of results. In this paper, we propose a novel method for homography estimation. We first estimate the projective transformation between the reference image and the target image at a coarse level and then refine the estimated homography at the fine level. Unlike approaches that use simple CNNs or global correlations to search correspondences, we add self- and cross-attention layers in the transformer to enhance the feature correlations. The experiments show that our method significantly outperforms the existing solutions in challenging large-baseline scenes.

Abstract Homography estimation serves an important role in many computer vision tasks. Depending heavily on hand‐craft feature quality, traditional methods degenerate sharply in scenes with low texture. Existing deep homography methods can handle the low‐texture problem but are not robust for scenes with low overlap rates and/or illumination changes. This paper proposes a novel unsupervised homography estimation method that can simultaneously handle such low overlap and illumination change. Specifically, a powerful module, named global transformer contextual encoder (GTCE) is first designed, together with a correlation encoder to effectively aggregate global contextual information and reduce matching ambiguity between feature maps. Moreover, a hybrid photo‐perceptual loss for unsupervised homography estimation is proposed. The proposed loss function considers alignment information on both pixel level and perceptual level thus helping this network to be more adaptive to various scenes, including normal cases and illumination change cases. The results of extensive experiments on synthetic and real‐world datasets demonstrate the superiority of this proposed method over current state‐of‐the‐art solutions especially on challenging scenes with low overlap rates, repetitive patterns and illumination changes.

Abstract The accuracy of registration in image stitching task directly affects the performance of subsequent stages. Traditional registration methods rely heavily on the quality of the features when calculating the homography matrix, resulting in alignment failures in low‐texture or low‐overlap scenes due to extracting insufficient features. On the other hand, existing DNN‐based methods for homography estimation are more robust in multiple scenes but previous work usually employs an overly simple convolutional network structure to directly regress the homography, ignoring the redundant information contained in the feature maps so that their prediction accuracy is inferior to the traditional methods in simple scenarios. To overcome the disadvantages of the two methods, a Multi‐scale structure is proposed to extract feature maps at three scales and design two modules to handle the matrix prediction respectively. The DFA‐T module analyzes semantic information on the high‐level features to accomplish coarse‐grained alignment while the Contextual Correlation module on the bottom level to accomplish more accurate alignment. Experiments demonstrate that this method provides more accurate alignment results than the existing state‐of‐the‐art DNN‐based methods and outperforms traditional algorithms with more stable results in some extreme scenarios.

Both traditional and deep learning-based methods have limitations in extracting statistical features from Polarimetric Synthetic Aperture Radar (PolSAR) images that contain regions with different levels of heterogeneity. To address this issue, we present a Statistical-Texture feature Learning Network (STLNet) for PolSAR image classification. Our approach includes several strategies. Firstly, we propose a novelNth-order Statistical feature Learning (N-SL) module as the statistical modeling interface to be combined with the network. In addition, we propose a Multi-level high-order Statistical feature Learning (MSL) module based on theN-SL module to represent the statistical characteristics of PolSAR images. Secondly, we propose a Texture feature Learning (TL) module to explore the spatial relationships among pixels and supplement the learned statistical features. Experimental results on E-SAR and AIRSAR datasets demonstrate that the proposed MSL and TL modules can effectively improve classification performance. Furthermore, STLNet outperforms other networks of comparable size.

RGBT tracking can be accomplished by constructing Bayesian estimators that incorporate fusion prior distributions for the visible (RGB) and thermal (T) modalities. Such estimators enable the computation of a posterior distribution for the variables of interest to locate the target. Incorporating rich prior information can improve the performance of predictors. However, current RGBT trackers face limited fusion prior data. To mitigate this issue, we propose a novel tracker, BD$^{2}$Track, which employs a diffusion model. Firstly, this paper introduces a dumbbell diffusion model, and employ convolution networks and the dumbbell model to derive the fusion feature prior information from various index frames in the same tracking video sequence. Secondly, we propose a plug-and-play channel augmented joint learning strategy to derive the images prior distribution. This strategy not only homogeneously generates modality-relevant prior information but also increases the distance between positive and negative samples within the modality, while reducing the distance between modalities during fusion. Results demonstrate promising performance in the GTOT, RGBT234, LasHeR, and VTUAV-ST datasets, surpassing other state-of-the-art trackers.

The varying polarimetric orientation angles (POAs) result in scattering diversity, leading to ambiguity in the interpretation of polarimetric synthetic aperture radar (PolSAR) images. Exploring the scattering characteristics in the polarimetric rotation domain (PRD) and the complementary features can help overcome the ambiguity. To address this, we propose a novel PolSAR image classification algorithm called learning scattering similarity and texture-based attention with convolutional neural networks (LSTCNNs). Three strategies are included in the proposed method. First, a pixel-level scattering similarity learning (SSL) module is proposed to analyze the scattering components of radar targets by learning the mapping from PolSAR data in the PRD to typical scattering models, with rotation angles as learnable parameters to utilize scattering diversity and avoid ambiguity. Second, a neighborhood-level texture-based attention (TA) module is proposed to learn the spatially enhanced features of PolSAR images, with the attention module design guided by the physical meaning of texture and consideration of channel and position importance. Finally, the proposed LSTCNN, which includes the SSL module, the TA module, and the classification module, combines pixel-level scattering features in the PRD and neighborhood-level texture features to increase the discriminability of features. The experimental results on three PolSAR images acquired by airborne SAR (AIRSAR) and experimental SAR (E-SAR) demonstrate the robustness and excellence of LSTCNN.

Convolutional neural networks(CNNs) currently dominate the field of computer vision, where the pooling layer plays an important role in reducing computational effort and avoiding overfitting. However, the commonly used methods do not design the pooling layer from the perspective of frequency. In this paper, we propose a Lifting Scheme-based frequency-dividing downsampling framework and describe a pooling layer called frequency-dividing pooling (FDP). The two branches of the Lifting Scheme process the images by frequency, which not only enhances the interpretability of the neural network but also improves the classification accuracy of the neural network. We conduct experiments on three standard datasets and the results all demonstrate that our proposed FDP is effective.

Image stitching aims to warp and align two or more images with overlapping areas to generate a panorama with a larger field of view. Due to the wide baseline or the abrupt depth changes caused by the foreground objects, the phenomenon that adjacent pixels or regions in one image are not adjacent in another image happens. It is difficult to avoid severe parallax artifacts and get good alignment results when stitching such images. In this paper, focusing on the images with large parallax, we design a multi-plane alignment algorithm guided by the disparity map. The concept of average tolerable parallax is proposed to help distinguish one background plane and multiple foreground objects from the image, which are robust to small parallax. In order to obtain the reliable disparity map of the image pairs got from any camera with unknown camera parameters, we propose an optimal homography estimation method based on the relative projection biases. This helps to satisfy the common epipolar line constraint when applying stereo matching modules. Experimental results demonstrate that our algorithm provides accurate stitching results on images with large parallax, and outperforms other existing state-of-the-art methods both qualitatively and quantitatively.

Synthetic aperture radar (SAR) images inherently present random and complex spatial patterns, which makes the land-cover classification from SAR images a challenging task. A convolutional neural network (CNN) has been applied to the land-cover classification. However, the statistical properties of an SAR image have not yet been explicitly considered by CNN for feature extraction. To address this problem, this letter presents a statistical CNN (SCNN) for land-cover classification from SAR images, which enables the representation of learning and statistical analysis to be implemented with a unified framework. In the proposed SCNN, the distribution of mid-level primitive features, extracted by representation learning, is characterized by their first- and second-order statistics. These statistics are used to fit the land-cover representations, which encode the statistical properties of the SAR image in the feature space. Experiments on the TerraSAR-X data demonstrate that the SCNN is effective and efficient for the land-cover classification from SAR images.

Deep Learning has shown great success in many fields, however, transferring this potential to remote sensing imagery interpretation is still a challenging task due to the special data properties, e.g., low Signal-to-Noise Ratio (SNR), high variation, etc. In this work, a lifting scheme based deep model is presented for remote sensing imagery classification. The main idea underlying this scheme is that, an innovative strategy is adopted to decompose the input image into two compact and low-resolution components, and these components are then fed into a standard Convolutional Neural Network (CNN) for classification task. More precisely, (1) one decomposed component is devoted to enhancing the latent patterns and simultaneously attenuating the random variation in the input, and (2) the other component is used to capture the local structural information in the input. The experimental results show that the lifting deep model is computationally efficient and has promising potential, improving the classification accuracy by about 5.7% and obtaining 2.69 x speed-up compared with the counterpart CNN.

How to exploit the rich information contained in Polarimetric Synthetic Aperture Radar (PolSAR) data, has recently gained much attention for PolSAR image interpretation via Deep Learning. In this paper, a polarimetric phase difference aided approach is presented for PolSAR image classification. The polarimetric phase differences conveyed by the off-diagonal elements, as well as the diagonal components in the coherence matrix, are extracted to form a 6-D target vector, i.e., the input to a deep model includes 6 channels. The experimental results on benchmark PolSAR data indicate that the polarimetric phase difference is indeed information bearing, moreover, the proposed strategy can be flexibly implemented by current deep learning framework without modification.

Convolutional Neural Network (CNN) has shown great potential for structural information extraction, however, Synthetic Aperture Radar (SAR) image interpretation via CNN, is still a challenging task due to the pixel- to- pixel variation nature of SAR signal. In this paper, a pattern strengthened deep model for SAR image classification is presented. The main idea underlying this method is to strengthen (or sharpen) the salient patterns contained in SAR image while attenuate the random variation, so as to leverage the enormous potential of deep model. To achieve this goal, max pooling is elaborately designed to form a pattern strengthened deep model. The experimental results on two SAR datasets show that the proposed method increases the classification accuracy by 4.91% and 4.64% compared with the typical CNN, and indicate our innovative attempt is promising for SAR image classification in the case of limited training data.

Image classification performance depends on the understanding of image features and classifier selection. Owing to the special imaging mechanism, achieving precise classification for remote sensing imagery is still quite challenging. In this paper, a fully convolutional network with polarimetric manifold, is proposed for Synthetic Aperture Radar (SAR) image classification. First, the polarimetric features are extracted to describe the target information; then the feature points in high-dimension are mapped to low-dimension through the manifold structure. In this way, the effect of single manifold is equal to that of multi -layer convolution. The experimental results on SAR image data indicate that the presented manifold network can effectively separate the polarimetric features and improve the classification accuracy.

The conventional studies on different types of remote sensing (RS) images classifications are conducted separately. Thanks to the powerful potential of deep learning to automatically learn features from data, exploring a unified method is possible. Moreover, recent research shows that sparse and low-rank representations can convey valuable information for patterns classification. Therefore, this paper presents a two-stream heterogeneous RS images unified interpretation network (HRSIUI-Net). One stream is to transfer the pre-trained fully convolutional network to learn deep multi-scale spatial features of RS data. The other stream is to employ a subspace learning based on graph embedding to learn the sparse and low-rank subspace representations of high-dimensional features. And then, two streams of learned subspace features are integrated for classification combined with an SVM. The experimental results on two typical RS data indicate that HRSIUI-Net can achieve competitive performance.

Time series polarimetric SAR image classification relies on learned understanding of how the set of pixels in an image relate by relative position and how the information of different dates in a time series change as time goes on. In this paper, we firstly integrate the incoherent information in the spatial scale and the coherent information in the temporal scale to form the feature for time series polarimetric SAR images. Then we take advantage of the fully convolutional network (FCN) to make end-to-end, pixels-to-pixels 3-dimensional training, on which base we sparse the connection structure of deep learning network using low rank tensor decomposition to reduce the computational complexity of convolutional layers. Experiment results on a real polarimetric SAR data set preliminarily show the effectiveness of our presented approach.

Deep-learning-based methods often suffer from insufficient training samples when they are directly used in the task of Synthetical Aperture Radar (SAR) images classification, which in turn leads to poor performance. To alleviate this problem, this paper presents a feature-fused approach, in which several statistical features of SAR images are extracted and integrated into the first layer of a typical Convolutional Neural Networks (CNNs). Since SAR images exhibit evidently statistical properties, those statistical features, which are characterized by non-linearity and cannot be adaptively learnt by CNNs in the first layer, can be thought of as prior knowledge to facilitate performance enhancement. Experiments conducted on real TerraSAR-X dataset demonstrate the effectiveness of the proposed method, and the classification accuracy is improved by about 2%.

We propose a novel statistical distribution texton (s-texton) feature for synthetic aperture radar (SAR) image classification. Motivated by the traditional texton feature, the framework of texture analysis, and the importance of statistical distribution in SAR images, the s-texton feature is developed based on the idea that parameter estimation of the statistical distribution can replace the filtering operation in the traditional texture analysis of SAR images. In the process of extracting the s-texton feature, several strategies are adopted, including pre-processing, spatial gridding, parameter estimation, texton clustering, and histogram statistics. Experimental results on TerraSAR data demonstrate the effectiveness of the proposed s-texton feature.

In this paper, a classification method based on multi-layer network and transfer learning has been developed for synthetic aperture radar (SAR) images inspired by recent successful deep learning methods. Multi-layer network has excellent performance in the classification of optical images, while its application for SAR images is restricted by the limited quantity of SAR imagery training data. Given this, transfer learning has been introduced into the classification of a small number of SAR images. Firstly, we use CIFAR-10 dataset to train a multi-layer network in order for an extraction of the mid-level representation, and then we utilize the intermediate layers of the network trained before to target SAR datasets at which the mid-level representation obtained can be used to train adaptive layers. The classification algorithm has been tested on a TerraSAR dataset and the results are more convincing and show greater potential for SAR image classification.

In this paper, a generative model based on methods for classification of time series of SAR images is proposed. For time series image classification, the feature expression of single image is based on parameters and simple texture features, which can't characterize the image well in some cases. Given this, we propose a generative model-LDA topic model on nonlinear compressed sensing method to better characterize the image with the latent semantic learning instead of the original features. Due to the complexity and uncertainty of the model parameters and the sparsity caused by the process of modeling, we further introduce Compressed Sensing (CS) to encode the model parameters for the distinction and stability. Experiments on the first batch of polarimetric SAR data and ESAR data demonstrate the effectiveness of the proposed LDA topic model. And the presented approaches good performance is proved by experiments on classification of time series of SAR images.

This letter presents a novel approach to learning the correct sampling positions by introducing a particle filter. A filtering, labeling, and statistics framework we previously proposed is applied to construct a complete texture descriptor named particle filter sample texton (PFST) feature for the classification of synthetic aperture radar (SAR) images. First, the gray values of the key points tracked by a particle filter in the local image patch are concatenated into a vector. Second, the vectors are labeled using a texton dictionary clustered from the training images. Finally, the histogram statistics is performed on these labels to generate the feature vectors for classification. The proposed method is more robust in terms of speckle noise and extremely low signal-to-noise ratio than those of the existing fixed-point and random sampling methods that play a significant role in popular binary textural descriptors. The experiments conducted on the TerraSAR image present evidence that the key points tracked by the particle filter effectively preserve the texture information, and the PFST feature performs best in the extreme situations.

This letter introduces a new image descriptor named Local Topographic Shape Pattern (LTSP) for texture description by relying on complete shape-based image representation. Firstly, a texture image is decomposed into a tree of shapes, i.e., topographic map, to obtain a multi-scale and contrast-invariant representation. Secondly, the shape of each node in the tree is handled with four shape patterns which are proposed as spatial codes of the shape. Finally, statistical histograms of shape patterns are used as texture descriptors. Contrast experiments of the proposed method show satisfactory performance in terms of both texture retrieval and classification tasks when applied to the Brodatz and UMD texture datasets.

In this paper, an algorithm applied for road extraction on SAR image is proposed, which is based on a multi-scale linear feature detector and beamlet framework, and then a quadratic kernel is introduced to offer optimal representation for the circle roads, aiming at improving the extraction quality. Firstly, a multi-scale pyramid is built on the input image and at each level the image is subdivided into a series of dyadic squares that constructs a quadtree. Then the multi-scale linear feature detector and beamlet are employed to compute pixels' responses. Finally, a quadratic kernel for non-linear candidates is introduced and adaptively selects the generating direction of segments. Experiments on TerraSAR images prove that the proposed approach significantly improves the extraction quality and performance when compared to several methods.

In this paper, a new Attributed Scattering Center(ASC) feature extraction model is proposed. Together with normalization procedure, optimization of amplitude and the length of scattering center feature extraction, we can get a fine estimation of ASC parameter. The image reconstruction experiment demonstrates that with fewer scattering center can we get a satisfied description of SAR image. Moreover, we also do classifaication experiments on TerraSAR-X data base, the result demonstrate that KNN classification method with ASC feature can obtain a better result than GLGM and GMRF. In this way the usage of ACS is exterded.

In this paper, an algorithm applied for building area extraction on SAR image is proposed, which is based on conditional random model, then a boundary prior relation is introduced to strengthen the description of prior item around the edge of building area, aiming at improving the classification performance nearby the boundary lines encompass building area. Firstly, pre-segmentation and boundary lines extraction can be accomplished respectively rely on mean shift algorithm and ratio of average edge detection. After that a combination term of the distances between the boundary lines and pixels around them and the pixels' label information can help to improve the prior item in CRF and build the boundary prior-CRF model. Finally, several experimental results on TerraSAR-X images prove that the proposed approach significantly improves the extraction accuracy and classification performance when compared to CRF.

This letter proposed a Part-based CFAR Model for object detection of power tower on high-resolution SAR images. Firstly, Part-based Model is used to describe the structure feature of the target, then Compressing Sensing approach is added to reduce the speckle by means of rebuilding background clutter, next, CFAR method is used to extract local shape and scale parameters, at last, Part-based CFAR Model combines these procedures together to form the finally algorithm, not only includes the distribution features, but also considers the structure relationship in the proposed approach. The algorithm is tested on TerraSAR-X data set with the resolution of 1m and 3m. Experiments show that unlike the CFAR method can only gives the high-light points of the targets; Part-based CFAR Model illuminates the target and its local components by plotting the bounding boxes around them.

This work introduces an unsupervised classification framework based on ensemble partitioning for polarimetric synthetic aperture radar (PolSAR) data, which can automatically determine the number of categories. First, the PolSAR image is divided into patches by an over-segmentation method. Second, ensemble partitioning is performed on the patch based dataset to obtain an ensemble similarity matrix. Third, a self-tuning spectral clustering method is adopted to automatically find the number of categories and the classification results, which is finally smoothed by a Markov random field based method. The experimental results on PolSAR image show the effectiveness of this unsupervised classification method.

This paper presents a frame for classifying polarimetric synthetic aperture radar (PolSAR) data. The frame is based on the combination of wavelet polarization information, textons, and sparse coding. Polarimetric synthesis unites with the discrete wavelet frame to obtain wavelet polarization variance through the calculation of the wavelet variance in the space of polarization states. The K-means cluster algorithm is implemented to cluster the wavelet polarization variance vectors of the training samples for the purpose of constructing a texton dictionary. A patch, in which all the wavelet polarization variance vectors match those in the texton dictionary, is used to obtain a statistical histogram. Sparse coding is applied to describe the histogram feature and generate a new texture feature called sparse coding of a wavelet polarization texton. Finally, support vector machine is used for the classification. All experiments are carried out on five sets of PolSAR data. The experimental results confirm that the proposed method effectively classifies PolSAR data.

This paper, we propose a novel over-segmentation method Feature Geometry Space Fusion (FGSF) for polarimetric SAR (POLSAR) data classification, which uses the polarimetric feature and geometric feature. In order to exam its performance, experiments on the data acquired by AIRSAR show that the over segment regions segmented by FGSF method performs better than meanshift when used for classification.

In this paper, the algorithm of K-SVD learning dictionary is applied to target decomposition for polarimetric SAR (PolSAR) images. This algorithm can obtain a set of bases self-adaptively according to the data on each channel of PolSAR, to make polarimetric data become more differentiated on this set of bases. Experiments on the data acquired through polarization SAR equipment developed by China for the first time show that features decomposed through K-SVD algorithm perform better than features based on the physical mechanism of PolSAR when used for classification.

This paper presents a novel approach for super-resolution (SR) reconstruction in Synthetic Aperture Radar (SAR), based on multi-dictionary. In comparison with conventional SR via sparse representation, the algorithm combines the classification with sparse representation. After classifying the training image, we jointly train the low and high resolution dictionaries for each class. And then, the image patches are reconstructed according to different dictionaries, which are chosen in conformity with the class of the image patches. The effectiveness of this method is demonstrated on Terra-SAR datasets.

This letter presents a supervised classification method for synthetic aperture radar (SAR) images based on multiscale region connection calculus (RCC) and conditional random fields (CRF). Using this method, first, a SAR image is oversegmented into multisuperpixels via the image pyramid. We then use the multiscale RCC model to describe the spatial logic relationships among these superpixels. To complete the process, multiscale RCC relationships are learned and reasoned under the CRF reasoning framework. This method employs iteration strategy for CRF reasoning to get better details in the classification results as well. We illustrate the proposed method by experiments conducted on DLR ESAR image. The results reveal efficient performance.

This paper describes a framework for multi-scale gray level analysis of images. It defines scales based on gray levels and organizes the basic “atoms” with a topographic map. The aim of this approach is to separate a large number of pixels concentrating in a narrow range of gray values. The main advantage of the methodology is that it allows manipulating pixels according to gray levels and spatial relations simultaneously. We apply it to histogram modification of Synthetic Aperture Radar (SAR) images. The experiments on displaying and classification prove the superiority of the approach.

In this paper, we present a fast approach to obtain semantic scene segmentation with high precision. We employ a two-stage classifier to label all image pixels. First, we use the regularized logistic regression to combine different appearance-based features and the improved spatial layout of labeling information. In the second stage, we incorporate the local, regional and global cues into a conditional random field model to provide a final segmentation, and a fast max-margin training method is employed to learn the parameters of the model quickly. The comparison experiments on four multi-class image segmentation databases show that our approach can achieve comparable semantic segmentation results and work faster than that of the state-of-the-art approaches.

Inspired by Weber's Law, this paper proposes a simple, yet very powerful and robust local descriptor, called the Weber Local Descriptor (WLD). It is based on the fact that human perception of a pattern depends not only on the change of a stimulus (such as sound, lighting) but also on the original intensity of the stimulus. Specifically, WLD consists of two components: differential excitation and orientation. The differential excitation component is a function of the ratio between two terms: One is the relative intensity differences of a current pixel against its neighbors, the other is the intensity of the current pixel. The orientation component is the gradient orientation of the current pixel. For a given image, we use the two components to construct a concatenated WLD histogram. Experimental results on the Brodatz and KTH-TIPS2-a texture databases show that WLD impressively outperforms the other widely used descriptors (e.g., Gabor and SIFT). In addition, experimental results on human face detection also show a promising performance comparable to the best known results on the MIT+CMU frontal face test set, the AR face data set, and the CMU profile test set.

In this paper, a Bayesian LBP operator is proposed. This operator is formulated in a novel Filtering, Labeling and Statistic (FLS) framework for texture descriptors. In the framework, the local labeling procedure, which is a part of many popular descriptors such as LBP, SIFT and VZ, can be modeled as a probability and optimization process. This enables the use of more reliable prior and likelihood information and reduces the sensitivity to noise. The BLBP operator pursues a label image, when given the filtered vector image, by maximizing the joint probability of two images under the criterion of MAP. The proposed approach is evaluated on texture retrieval schemes using entire Brodatz database. The result reveals BLBP operator’s efficient performance and FLS framework’s capability to in-depth analysis of the texture descriptors on a common background.

This letter presents a precise and rapid clustering method for synthetic aperture radar (SAR) images by embedding a Markov random field (MRF) model in the clustering space and using graph cuts (GCs) to search the optimal clusters for the data. The proposed method is optimal in the sense of maximum a posteriori (MAP). It automatically works in a two-loop way: an outer loop and an inner loop. The outer loop determines the cluster number using a pseudolikelihood information criterion based on MRF modeling, and the inner loop is designed in a ldquohardrdquo membership expectation-maximization (EM) style: in the E step, with fixed parameters, the optimal data clusters are rapidly searched under the criterion of MAP by the GC; and in the M step, the parameters are estimated using current data clusters as ldquohardrdquo membership obtained in the E step. The two steps are iterated until the inner loop converges. Experiments on both simulated and real SAR images test the performance of the algorithm.

This letter presents a new method for unsupervised urban area extraction from synthetic aperture radar (SAR) images based on the ffmax algorithm proposed by C. Gouinaud specially for acquiring urban areas in SPOT imagery. According to the statistical characteristics of urban areas, an adaptive and iterative method based on the low-level extraction given by the ffmax algorithm using a large window is proposed. Experimental results on real SAR images show that the proposed automatic method works quickly and can preserve the borders of urban areas as well as avoid the disturbance of other classes and the extractions of urban areas are reliable and precise

This letter presents a novel method of supervised multiresolution segmentation for synthetic aperture radar images. The method uses a region-based half-tree hierarchical Markov random field model for multiresolution segmentation. To form the region-based multilayer model, the watershed algorithm is employed at each resolution level independently. The nodes of a quadtree in the proposed model are defined as regions instead of pixels. The relationship over scale is studied, and the region-based upward and downward maximization of posterior marginal estimations are deduced. The experimental results for the segmentation of homogeneous areas prove the region-based model much better in terms of robustness to speckle and preservation of edges compared to the pixel-based hierarchical model and the Gibbs sampler with the single-resolution model