Synthetic aperture radar (SAR) ship image classification is of great significance in the field of marine ship monitoring. Extracting effective feature representation and constructing suitable classifier can fundamentally improve the accuracy of ship classification. At present, using distance metric learning (DML) algorithm to learn effective distance metrics for classifiers has been widely used in information retrieval and face recognition, but its ability to implement SAR ship image classification is still unknown. In this paper, we show the performance of 4 feature representations and 20 DML algorithms in SAR ship classification. Experimental results show that extracting effective feature representation is essential, and the DML algorithm has the ability to learn better distance metrics.
Knowledge about ship positions plays a critical role in a wide range of maritime applications. To improve the performance of ship detector in SAR image, an effective strategy is improving the signal-to-clutter ratio (SCR) before conducting detection. In this paper, we present a comparative study on methods of improving SCR, including power-law scaling (PLS), max-mean and max-median filter (MMF1 and MMF2), method of wavelet transform (TWT), traditional SPAN detector, reflection symmetric metric (RSM), scattering mechanism metric (SMM). The ability of SCR improvement to SAR image and ship detection performance associated with cell- averaging CFAR (CA-CFAR) of different methods are evaluated on two real SAR data.
In recent years, oil spill surveillance with space-borne synthetic aperture radar (SAR) has received unprecedented attention and has been gradually developed into a common technique for maritime environment protection. A typical SAR-based oil spill detection process consists of three steps: (1) dark-spot segmentation, (2) feature extraction, and (3) oil spill and look-alike discrimination. As a preliminary task in the oil spill detection process chain, dark-spot segmentation is a critical and fundamental step prior to feature extraction and classification, since its output has a direct impact on the two subsequent stages. The balance between the detection probability and false alarm probability has a vital impact on the performance of the entire detection system. Unfortunately, this problem has not drawn as much attention as the other two stages. A specific effort has been placed on dark-spot segmentation in single-pol SAR imagery. A combination of fine designed features, including gray features, geometric features, and textural features, is proposed to characterize the oil spill and seawater for improving the performance of dark-spot segmentation. In the proposed process chain, a histogram stretching transform is incorporated before the gray feature extraction to enhance the contrast between possible oil spills and water. A simple but effective multiple-level thresholding algorithm is developed to conduct a binary classification before the geometric feature extraction to obtain more accurate area features. A local binary pattern code is computed and assigned as the textural feature for a pixel to characterize the physical difference between oil spills and water. The experimental result confirms that the proposed fine designed feature combination outperforms existing approaches in both aspects of overall segmentation accuracy and the capability to balance detection probability and false alarm probability. It is a promising alternative that can be incorporated into existing oil spill detection systems to further improve system performance.
Ship classification in synthetic aperture radar (SAR) imagery has become a new hotspot in remote sensing community for its valuable potential in many maritime applications. Several kinds of ship features, such as geometric features, polarimetric features, and scattering features have been widely applied on ship classification tasks. Compared with polarimetric features and scattering features, which are subject to SAR parameters (e.g., sensor type, incidence angle, polarization, etc.) and environment factors (e.g., sea state, wind, wave, current, etc.), geometric features are relatively independent of SAR and environment factors, and easy to be extracted stably from SAR imagery. In this paper, the capability of geometric features to classify ships in SAR imagery with various resolution has been investigated. Firstly, the relationship between the geometric feature extraction accuracy and the SAR imagery resolution is analyzed. It shows that the minimum bounding rectangle (MBR) of ship can be extracted exactly in terms of absolute precision by the proposed automatic ship-sea segmentation method. Next, six simple but effective geometric features are extracted to build a ship representation for the subsequent classification task. These six geometric features are composed of length (f<sub>1</sub>), width (f<sub>2</sub>), area (f<sub>3</sub>), perimeter (f<sub>4</sub>), elongatedness (f<sub>5</sub>) and compactness (f<sub>6</sub>). Among them, two basic features, length (f<sub>1</sub>) and width (f<sub>2</sub>), are directly extracted based on the MBR of ship, the other four are derived from those two basic features. The capability of the utilized geometric features to classify ships are validated on two data set with different image resolutions. The results show that the performance of ship classification solely by geometric features is close to that obtained by the state-of-the-art methods, which obtained by a combination of multiple kinds of features, including scattering features and geometric features after a complex feature selection process.
Ship surveillance by remote sensing technology has become a valuable tool for protecting marine environments. In recent years, the successful launch of advanced synthetic aperture radar (SAR) sensors that have high resolution and multipolarimetric modes has enabled researchers to use SAR imagery for not only ship detection but also ship category recognition. A hierarchical ship detection and recognition scheme is proposed. The complementary information obtained from multipolarimetric modes is used to improve both the detection precision and the recognition accuracy. In the ship detection stage, a three-class fuzzy c-means clustering algorithm is used to calculate the segmenting threshold for prescreening ship candidates. To reduce the false alarm rate (FAR), we use a two-step discrimination strategy. In the first step, we fuse the detection results from multipolarimetric channels to reduce the speckle noise, ambiguities, sidelobes, and other sources of interference. In the second step, we use a binary classifier, which is trained with prior data collected on ships and nonships, to reduce the FAR even further. In the ship category recognition stage, we concatenate texture-based descriptors extracted from multiple polarmetric channels to construct a robust ship representation for category recognition. Furthermore, we construct and release a ship category database with real SAR data. We hope that it can be used to promote investigations of SAR ship recognition in the remote sensing and related academic communities. The proposed method is validated by a comprehensive experimental comparison to the state-of-the-art methods. The validation procedure showed that the proposed method outperforms all of the competing methods by about 5% and 15% in terms of ship detection and recognition, respectively.
A novel one-shot in-line digital holography based two-dimensional Hilbert demodulation is proposed. By weakening the
object wave compared with the reference wave and applying natural logarithmized operation on the in-line digital
hologram, the real part of object wave can be well extracted. Then utilizing two-dimensional Hilbert transform to
digitally realize π / 2 phase-shift makes it possible to reconstruct the object wave front from single-exposure in-line
digital hologram. Preliminary experimental results are presented to demonstrate the proposed method. This technique
can be used for real time imaging and monitoring moving objects.
Proc. SPIE. 6294, Infrared and Photoelectronic Imagers and Detector Devices II
KEYWORDS: Mathematical modeling, Signal to noise ratio, MATLAB, Cameras, Coded apertures, Algorithm development, Optimization (mathematics), Coded aperture imaging, Binary data, Simulation of CCA and DLA aggregates
Coded aperture imaging (CAI) has evolved as a standard technique for imaging high energy photon sources and has
found numerous applications. Coded aperture arrays (CAAs) are the most important devices in the applications of CAI.
In recent years, many approaches were presented to design optimum or near-optimum CAAs. Uniformly redundant
arrays (URAs) are the most successful CAAs for their cyclic autocorrelation consisting of a sequence of delta functions
on a flat sidelobe which can easily be subtracted when the object has been reconstructed. Unfortunately, the existing
methods can only be used to design URAs with limited number of array sizes and fixed autocorrelative sidelobe-to-peak
ratio. In this paper, we presented a method to design more flexible URAs by means of a global optimization algorithm
named DIRECT. By our approaches, we obtain various types of URAs including the filled URAs which can be
constructed by existing methods and the sparse URAs which never be constructed and mentioned by existing papers as
far as we know.
Three-dimensional imaging techniques are very attractive for many applications. We develop the basic principle of coded aperture imaging used in invisible imaging realm to visible imaging realm, propose a three-dimensional imaging method. The object is captured by a cameras array. Then captured photographs of the object are integrated into an image named coded image. Finally coded image is computationally decoded to obtain a series of longitudinal layered surface images of the object. For good reconstructed images fidelity, we make use of correlation decoding method. With the use of correlation decoding, the distribution of cameras in array is crucial for the quality of reconstructed images. We investigate some typical two-dimensional arrays, choose non-redundant array for its proper imaging property. Experiments have been done to test and verify the performance of the proposed method. We choose a simple discontinuous object. The object is composed of two digit models, digit "1" and "2". Two digit models are displaced from each other. The distance between them is 10cm. Cameras array includes 9 cameras arranged as non-redundant array. The object is placed at the center axis of the cameras array, face to face with the array. After capturing, photographs integrating, computational decoding etc. procedures, we obtain high-quality reconstructed images of digit "1" and "2". The results of experiments show that the proposed method is feasible.
A method using the phase-space representations, i.e. the ambiguity function or Wigner distribution function to compute the optical transfer function (OTF) for an optical system with circularly symmetrical pupils under polychromatic illumination is presented. The phase-space representations is a very convenient tool for display the optical transfer function with varying aberrations such as the longitudinal chromatic aberration and defocus in a single picture, and the monochromatic OTFs can be easily determined from these joint representations. The polychromatic OTFs are computed by synthesizing a suitable number of monochromatic OTFs weighted by the spectral distribution of source and the color sensitivity of the receiver at fixed wavelength. Since the ambiguity function or the Wigner distribution function can be previously obtained by optical method or digital computation, the computational efficiency is greatly improved compared with traditional method, in which every monochromatic OTF need to be determined along. We computed the polychromatic OTFs for an optical system with a clear circular pupil and an annular ring pupil in detail and show some primary applications of the computations in spatial filter designing for color-blur reduction.