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

Since high-resolution inverse synthetic radar (ISAR) images are significant in the analysis of moving targets, this paper, for the first time, proposes a novel method to compensate the uniform rotational motion by estimating the effective rotational velocity (ERV) in image domain. Specifically, we extract high-energy scatterers of sequent ISAR images by scale invariant feature transform (SIFT) in order to link them with the related ERV in the imaging panel. Then we adopt registering approach based on minimum Euclidian distance and Gaussian probabilistic density function to select scatterers with the mostly similar rotational aspect. After that, we estimate the ERV in a searching manner, and then employ it in the cancellation of quadratic-order phase items in the cross-range cells of the image. This method is validated quick and efficient by both simulated and real signal experiments.

Uniform rotational motion compensation for inverse synthetic aperture radar targets from image domain

According to the characteristics of SAR imaging, a SAR real-time imaging pipeline technology based on multicore DSP is proposed. The algorithm uses sub-aperture complex image stitching technology for real-time data processing, uses ECS operation to process the range direction of each subaperture data, and uses the Dechirp operation to process the azimuth direction. This technology is based on TI’s multi-core DSP C6678 chip to achieve a pipeline design, and coherently splices the processed sub-aperture image data in the complex image domain to obtain a full-resolution image of the recorded data. The algorithm is optimized for the KeystoneII hardware architecture of the chip to meet the performance requirements of real-time processing of SAR imaging. In this paper, the banding mode of SAR is verified, and the imaging results of the measured data test the effectiveness and real-time performance of the technology.

SAR real-time imaging pipeline technology based on multi-core DSP

Since the United States launched the first synthetic aperture radar satellite in 1978, it has attracted widespread attention because of its ability to obtain high-resolution images of almost every corner of the earth, all day, all weather, and regardless of national borders and politics. At present, spaceborne SAR has been widely used in civil and military applications. With the complexity of spaceborne SAR application scenarios, different imaging widths and imaging resolutions are required to deal with different imaging scenarios, imaging targets, and imaging tasks. For different imaging requirements, different imaging algorithms have been proposed, including strip imaging mode, spotlight imaging mode, scan imaging mode, etc. It is of great significance to integrate multiple imaging modes in a spaceborne SAR. Different imaging modes have different requirements for antenna beam pointing. And each imaging mode has the need to quickly switch beam pointing. The phased array antenna can realize the switching of beam pointing with different beam control codes, and the speed of beam direction switching depends on the calculation speed of beam control code. Therefore, the calculation strategy and the calculation speed of the beam control code have become the main problems which restricting the multi-mode imaging of spaceborne SAR. In this paper, a real-time antenna beam control method based on the computing framework of DSP, FPGA and CPU is proposed. According to different imaging modes, the corresponding beam control code calculation strategy is designed. The test results show that relying on the high-performance computing power of DSP and the logic control capabilities of FPGA and CPU, it perfectly meets the needs of quickly completing beam control code calculation to realize beam pointing switching in different imaging modes. This method makes it possible to realize multimode imaging of spaceborne SAR.

An antenna beam control method of spaceborne multi-mode imaging SAR based on real-time computing framework

With the development of digital radio frequency memory technology, the main-lobe deception jamming represented by interrupted-sampling repeater jamming (ISRJ) poses a severe challenge to radar. Traditional antijamming methods usually need to estimate the jamming parameters and have the risk of losing target information. For the above problems, this article proposes a deep neural network-based ISRJ recognition and antijamming target detection method which consists of four serial steps. First, the proposed method obtains the time-frequency image set of radar echoes by short-time Fourier transform (STFT). Second, a you-only-look-once (YOLO) model is used to detect the jammed echoes, and the positioning result is automatically corrected to avoid losing the target information. Third, the anti-ISRJ target ranging and velocity measurement datasets are constructed according to the positioning result. Finally, an anti-ISRJ target detection model based on the convolution neural network (CNN) is designed to extract features along different dimensions and obtain the range and velocity of the real targets. Experiments on simulated and measured datasets show that the proposed method has better antijamming detection performance than the traditional method, and does not need to estimate the jamming parameters.

Mengdao Xing

Papers

Uniform rotational motion compensation for inverse synthetic aperture radar targets from image domain

SAR real-time imaging pipeline technology based on multi-core DSP

An antenna beam control method of spaceborne multi-mode imaging SAR based on real-time computing framework

Deep Neural Network-Based Interrupted Sampling Deceptive Jamming Countermeasure Method

Man-made Target Detection in the Complicated Scene of High-resolution SAR: Man-made Target Detection in the Complicated Scene of High-resolution SAR