Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Обнаружение транспортных средств на изображениях загородных шоссе на основе метода Single shot multibox Detector

This chapter introduces the subject of statistical pattern recognition (SPR). It starts by considering how features are defined and emphasizes that the nearest neighbor algorithm achieves error rates comparable with those of an ideal Bayes’ classifier. The concepts of an optimal number of features, representativeness of the training data, and the need to avoid overfitting to the training data are stressed. The chapter shows that methods such as the support vector machine and artificial neural networks are subject to these same training limitations, although each has its advantages. For neural networks, the multilayer perceptron architecture and back-propagation algorithm are described. The chapter distinguishes between supervised and unsupervised learning, demonstrating the advantages of the latter and showing how methods such as clustering and principal components analysis fit into the SPR framework. The chapter also defines the receiver operating characteristic, which allows an optimum balance between false positives and false negatives to be achieved.

Statistical Pattern Recognition

Object detection in optical remote sensing images, being a fundamental but challenging problem in the field of aerial and satellite image analysis, plays an important role for a wide range of applications and is receiving significant attention in recent years. While enormous methods exist, a deep review of the literature concerning generic object detection is still lacking. This paper aims to provide a review of the recent progress in this field. Different from several previously published surveys that focus on a specific object class such as building and road, we concentrate on more generic object categories including, but are not limited to, road, building, tree, vehicle, ship, airport, urban-area. Covering about 270 publications we survey (1) template matching-based object detection methods, (2) knowledge-based object detection methods, (3) object-based image analysis (OBIA)-based object detection methods, (4) machine learning-based object detection methods, and (5) five publicly available datasets and three standard evaluation metrics. We also discuss the challenges of current studies and propose two promising research directions, namely deep learning-based feature representation and weakly supervised learning-based geospatial object detection. It is our hope that this survey will be beneficial for the researchers to have better understanding of this research field.

A survey on object detection in optical remote sensing images

In this paper, we examine the role of polarimetry in synthetic aperture radar (SAR) interferometry. We first propose a general formulation for vector wave interferometry that includes conventional scalar interferometry as a special case. Then, we show how polarimetric basis transformations can be introduced into SAR interferometry and applied to form interferograms between all possible linear combinations of polarization states. This allows us to reveal the strong polarization dependency of the interferometric coherence. We then solve the coherence optimization problem involving maximization of interferometric coherence and formulate a new coherent decomposition for polarimetric SAR interferometry that allows the separation of the effective phase centers of different scattering mechanisms. A simplified stochastic scattering model for an elevated forest canopy is introduced to demonstrate the effectiveness of the proposed algorithms. In this way, we demonstrate the importance of wave polarization for the physical interpretation of SAR interferograms. We investigate the potential of polarimetric SAR interferometry using results from the evaluation of fully polarimetric interferometric shuttle imaging radar (SIR)-C/X-SAR data collected during October 8-9, 1994, over the SE Baikal Lake Selenga delta region of Buriatia, Southeast Siberia, Russia.

Polarimetric SAR interferometry

Conventional sparsity-driven synthetic aperture radar (SAR) imagery proceeds via $\ell _{1}$ regularization, or named by least absolute shrinkage and selection operator (LASSO). However, followed by the enhanced sparse feature, structures of the scenes or targets of interest in weak scattering are easily lost. Therefore, it becomes difficult to make use of the high-resolution SAR data, and even high costs have been paid for the resolution. In this article, a novel structure-guaranteed SAR (SG-SAR) imaging algorithm is proposed by utilizing the morphology metric for the cluster feature of the scatterers of the scenes/targets of interest. By introducing structural priors in terms of morphology norm, the intended structure features can be highlighted via convex regularization. More specifically, to accommodate complicated scenarios or targets, the structure element (SE) in the morphology regularizer is designed to be spatially variant under structure tensor representation. Different from conventional convex optimizations, the proposed SG-SAR algorithm is solved under the alternating direction method of multipliers (ADMMs), which is flexible to incorporate with the super-resolution imagery. In such cases, both sparse and structural features can be simultaneously enhanced, even with limited measurements and in a low signal-to-noise ratio (SNR). Superior convergence and robustness can be guaranteed. Moreover, a grouping mask scheme is used to accommodate the complex-valued SAR data. Finally, both simulated and measured SAR data are applied for the validation. Comparisons with the conventions are performed in terms of phase transition analysis, so as to verify the superiority of the proposed algorithm both qualitatively and quantitatively.

Structure-Guaranteed SAR Imagery via Spatially-Variant Morphology Regularization in ADMM Manner

To improve the range resolution of stepped-frequency SAR system, frequency difference drift should be estimated using the raw data as accurately as possible. This paper proposes a new method to estimate the accurate frequency difference drift base on frequency band synthesis. In this method, the echoes of different subband signals will be compressed firstly so as to find out point targets with strong energy. Secondly, two phase functions are constructed by a pre-set frequency difference and multiplied to the echo signals of corresponding subband respectively; the interferometry phases are then obtained from the shifted subband signals. Finally a new normalized signal vector can be constructed from the interferometry phases; and the accurate frequency difference drift can be obtained by performing FFT to the vector. The required accuracy and the implementation strategies of this method are discussed, and the effectiveness and accuracy of this method are shown by some real SAR data results. (6 pages)

A new estimation method for SAR frequency difference drift base on frequency band synthesis

This paper presents a new method for estimating the baseline error of an azimuth multi-channel SAR antenna in the SAR image domain. In this paper, the expressions of the image domain of multi-channel SAR signals with azimuth baseline errors are derived. The covariance matrix of the image domain signals is obtained by using the joint pixel method. Finally, the least-squares method of image domain is deduced to estimate the azimuth baseline of multi-channel SAR error. Simulation experiments verify the effectiveness of the proposed method.

An Image-Domain Baseline Error Estimation Method for Azimuth Multi-Channel Sar

The ground-Cartesian factorized back-projection (G-CFBP) is an efficient time-domain processing algorithm without image interpolation and can realize accurate imaging for curved trajectory synthetic aperture radar (SAR). Its superiority shows good potential in airborne circular SAR (CSAR) imaging. However, the motion compensation (MoCo) based on ground Cartesian back-projection (GCBP) in the airborne CSAR is still a challenge. There are two main problems: one is that the existence of the image spectrum aliasing makes the processing of the phase error estimation inaccurate; the other is that the mapping relationship of the phase error between the image spectrum domain and the azimuth time domain still needs to be studied within GCBP processing chain. To tackle the above two problems, a novel MoCo method applicable to the GCBP algorithm is proposed and can be mainly divided into two steps: the first step is to remove the subaperture image spectrum aliasing by a spectrum compression operation; the second step is to establish an analytical phase error structure, which includes an autoselection criterion of the effective support region for GCBP image. The first step ensures the accuracy of the phase error estimation, and the second step establishes the inverse-mapping relationship of the phase error between the image spectrum and azimuth time. These two procedures are both vital in improving the accuracy and robustness of the GCBP-based MoCo for the CSAR imaging. The processed results of simulated and real data are provided to verify the effectiveness of the proposed method.

A Novel Motion Compensation Method Applicable to Ground Cartesian Back-Projection Algorithm for Airborne Circular SAR

In the field of image fusion, spatial detail blurring and color distortion appear in synthetic aperture radar (SAR) images and multispectral (MS) during the traditional fusion process due to the difference in sensor imaging mechanisms. To solve this problem, this paper proposes a fusion method for SAR images and MS images based on a convolutional neural network. In order to make use of the spatial information and different scale feature information of high-resolution SAR image, a dual-channel feature extraction module is constructed to obtain a SAR image feature map. In addition, different from the common direct addition strategy, an attention-based feature fusion module is designed to achieve spectral fidelity of the fused images. In order to obtain better spectral and spatial retention ability of the network, an unsupervised joint loss function is designed to train the network. In this paper, the Sentinel 1 SAR images and Landsat 8 MS images are used as datasets for experiments. The experimental results show that the proposed algorithm has better performance in quantitative and visual representation when compared with traditional fusion methods and deep learning algorithms.

/pdf/dafcnn-a-dual-channel-feature-extraction-and-attention-2cyfub8r.pdf

Mengdao Xing

Papers

Structure-Guaranteed SAR Imagery via Spatially-Variant Morphology Regularization in ADMM Manner

A new estimation method for SAR frequency difference drift base on frequency band synthesis

An Image-Domain Baseline Error Estimation Method for Azimuth Multi-Channel Sar

A Novel Motion Compensation Method Applicable to Ground Cartesian Back-Projection Algorithm for Airborne Circular SAR

DAFCNN: A Dual-Channel Feature Extraction and Attention Feature Fusion Convolution Neural Network for SAR Image and MS Image Fusion