In this paper, we proposed a novel two-stage model that works in a hierarchical architecture to detect airports on synthetic aperture radar (SAR) images. In the first rough stage, an improved line segment detector (LSD) which tackles the line segments disconnection problem in SAR images is used to extract candidate airport regions coarsely. In the second fine stage, a well-trained Faster R-CNN with residual networks (ResNet) implementation (called ResFaster RCNN) is applied to each candidate regions to discriminate airport and non-airport regions and locate the airport much more precisely. Transfer learning and some improvement measures are applied to the networks. With this two-stage model, we can not only use both low-level and high-level features of airports synthetically but also avoid the information loss problem caused by resizing when large scene images are input into the networks. Experiments on large-scale SAR images have proved that the proposed method can reach a detection rate of 95% with low false alarm rate and show a great enhancement over other existing methods.
As for multi-channel SAR, the spectrum of the clutter is spatially-temporally coupled, and the echo of the moving target
is chirp signal. A novel method based on STAP and FrFT is proposed in this paper, which is used for moving target
detection and parameter estimation. Two steps are used for fast target detection : the coarse detection in low range
resolution and parameter estimation for the specific data where the moving target appears. This paper discusses the
principle of frequency STAP for clutter suppression firstly, and subsequently infers that the signal after clutter
suppression is chirp signal. Then FrFT is introduced to estimate the parameters of the output signal, which can be used to
estimate the parameters of the moving target. Finally, the process of the proposed method is introduced. Matching
function is constructed to compensate the phase deviation caused by movement and focus the moving target. The
effectiveness of the proposed method is validated by the simulation.
In this paper, a novel three-dimensional imaging algorithm of downward-looking linear array SAR is presented. To improve the resolution, multiple signal classification (MUSIC) algorithm has been used. However, since the scattering centers are always correlated in real SAR system, the estimated covariance matrix becomes singular. To address the problem, a three-dimensional spatial smoothing method is proposed in this paper to restore the singular covariance matrix to a full-rank one. The three-dimensional signal matrix can be divided into a set of orthogonal three-dimensional subspaces. The main idea of the method is based on extracting the array correlation matrix as the average of all correlation matrices from the subspaces. In addition, the spectral height of the peaks contains no information with regard to the scattering intensity of the different scattering centers, thus it is difficulty to reconstruct the backscattering information. The least square strategy is used to estimate the amplitude of the scattering center in this paper. The above results of the theoretical analysis are verified by 3-D scene simulations and experiments on real data.
A method for target classification in synthetic aperture radar (SAR) images is proposed. The samples are first mapped into a high-dimensional feature space in which samples from the same class are assumed to span a linear subspace. Then, any new sample can be uniquely represented by the training samples within given constraint. The conventional methods suggest searching the sparest representations with ℓ1-norm (or ℓ0) minimization constraint. However, these methods are computationally expensive due to optimizing nondifferential objective function. To improve the performance while reducing the computational consumption, a simple yet effective classification scheme called kernel linear representation (KLR) is presented. Different from the previous works, KLR limits the feasible set of representations with a much weaker constraint, ℓ2-norm minimization. Since, KLR can be solved in closed form there is no need to perform the ℓ1-minimization, and hence the calculation burden has been lessened. Meanwhile, the classification accuracy has been improved due to the relaxation of the constraint. Extensive experiments on a real SAR dataset demonstrate that the proposed method outperforms the kernel sparse models as well as the previous works performed on SAR target recognition.
An unsupervised classification method combining Principal Component Analysis (PCA) and Gaussian Mixture Model for hyperspectral image is proposed in this paper. It is based on the property that lower dimensional linear projections of high dimensional data sets have the tendency to be Gaussian, or a combination of Gaussian distributions as the dimension increases. The spectral dimensionality of the data is first reduced by a PCA linear projection; then the transformed data is modeled by a Gaussian mixture models, the parameters of the model are estimated using the Expectation-Maximimization (EM) algorithm in merge operations and the number of components is automatically selected based on Bayesian Information Criterion (BIC); finally the data after PCA transform is classified according to the mixture model. Applying the method to Push-broom Hyperspectral Imager (PHI) data shows that the method is quite effective without any a prior information.
In this paper, we describe some new methods to perform color CCD camera calibration in virtual studio. Our work is based on the pinhole camera model with lens radial distortion. Generally, the calibration problem consists of two steps: control points extraction from images and calibration algorithm implementation. For control point extraction, the usual way is relatively simple and can be easily done. But because of the poor quality of grabbed images and the lens distortion, the usual method has two main disadvantages in our virtual studio application. Hence we present an effective and robust method to perform control points extraction. Generally, the calibration experiments are implemented by Tsai's two-stage algorithm. But in our experiments, we find that the calibration problem is ill-conditioned, for the result is very sensible to tiny changes of calibration points' coordinates. To solve this problem, we present a new strategy to improve the accuracy of calibration results of focal length and depth. Then we obtain a focal length look-up-table (LUT) relative to the lens zoom rings. In virtual studio application, when the real camera pans or tilts, the background picture lags behind the foreground picture. Then a motion compensation model is presented to compensate the difference between the optical center and the real rotation center to overcome the problem.
A target classification algorithm in high resolution and single polarimetric SAR image is presented in this paper. First, a new RCS reconstruction filter base don the modified correlated neighborhood model is used for target and shadow detection. By nonlinear iterated process, the whole image can be classified to 'background', 'shadow' and 'targets'. Secondly, a series of morphological operators are used to terrain filtering and edge extraction, and then through modified Hough Transform technology and the measure of slimming lines, line structures of man-made clutters in 'target' class are linked and grouped. Finally, with spatial association mode, the targets we are interested in are classified.
With reference to the air target detection of ultra- wideband(UWB)/impulse radar, we discussed transient signal processing techniques. In weak UWB signal detection, wavelet transforms and high order spectrum estimation techniques were preferred. In target characteristic analysis, two algorithms of impulse response deconvolution, MCGM and DPREM, were presented. In this paper, a time domain bispectrum estimation algorithm was used to analyze target impulse response, which could estimate accurately local scattering distribution of complex target. A free field IR experimental system was used which was laid out in an anechoic chamber. With this system, we measured the response of several target models and a scale aircraft.