Search

Since the existing target orientation estimation methods for laser radars are not robust to the occlusion,a novel 3D orientation estimation method was proposed.The principles of the existing methods and the intrinsic characters of the surface structures for ground targets were analyzed carefully,then the feature of Projection Density Entropy(PDE) was introduced,and the relationship between PDE and target orientation was analyzed.Finally,a 3D orientation estimation method was formulated by iteratively rotating and projecting the target point cloud onto the coordinate planes together with PDE calculation.Comparative experiments were performed on the dataset containing point clouds of 25 ground targets,the performance in self-occlusion,occlusion and the noise were investigated,and the parameter tuning was discussed.The results show that the estimation error of this method is below 3° in self-occlusion,and also no more than 10 ° even in 80% occlusion.It is clear that the proposed method is robust to occlusion and converges quickly,thus is capable of estimating the 3D orientation of ladar targets in complex scenes.

A lossless compression algorithm based on distributed source coding was proposed to compress the airborne hyperspectral data effectively.In order to make full use of the spectral correlation of hyperspectral images,a multi-band prediction scheme was introduced to acquire the prediction values of the current block and to reduce the maximal absolute value of prediction error effectively.Then,by using the maximal absolute value to determine the coset index of pixels belonging to the current block,the lossless compression of hyperspectral images was realized by transmitting the coset index of the current block instead of its prediction error.Experimental results on hyperspectral images acquired by Airborne Visible Infrared Imaging Spectrometer(AVIRIS) show that the proposed algorithm can offer both high compression performance and low encoder complexity compared with those existing classical algorithms,which is available for on-board compression of hyperspectral images.

A lossless compression algorithm based on contents was proposed for hyperspectral images.An adaptive band selection algorithm was introduced to reduce the dimensionality of hyperspectral images,and a C-means algorithm was used to classify the spectral vectors resulting from dimensionality reduction unsupervisedly.Then,the reverse monotonic ordering method was taken to determine the prediction ordering,hyperspectral images were divided into groups adaptively according to the correlation between each adjacent bands,and the scheme of multi-band linear prediction was used to eliminate the spectral redundancy of the identical class.For each class,partial pixels within this class were selected to train optimal predictive coefficients,and predictive errors were compressed in lossless by JPEG-LS standard.Experiments were performed for the hyperspectral images acquired by an Airborne Visible/Infrared Imaging Spectrometer(AVIRIS) and an Operational Modular Imaging Spectrometer(OMIS).Experiental results show that the average compression ratio of the proposed algorithm can be improved about 0.11 and 0.7 respectively as compared with above algorithms without classification prediction.

The paper proposed the new hyperspectral image compression algorithm by using block compressive sensing and inter-band prediction error. Because of making the best use of the intense spectral correlation between adjacent bands, we can obtain the higher reconstruction quality and lower complexity. In this method, we firstly estimate the prediction parameters between adjacent bands by using least squares method. Then each band is measured by the same random projection matrix independently. Then the prediction parameters and random measurements are transmitted to the decoder. At last, we exploit a new restore process and prediction information to reconstruct the hyperspectral images. The prediction information is computed based on the previous restore adjacent band and the prediction parameters. Several experimental results demonstrate that the proposed algorithm is obviously better than the traditional compressive sensing, and has a very low complexity. In addition, our algorithm is easy to realize in practice.

For hyperspectral image classification,a Support Vector Machine(SVM) algorithm with composite kernels was presented to fuse both the spectral information and spatial information of the image.The algorithm adopts Principal Component Analysis(PCA) algorithm to extract the image feature and reduce the dimension for hyperspectral image,and uses the Virtual Dimension(VD) algorithm to estimate the Intrinsic Dimension(ID) of the image.Then,the remained number of Principal Components(PCs) was determined on the basis of the ID.Furthermore,spatial features were extracted by mathematical morphology from the remained PCs,and the Extended Morphological Profile(EMP) vector of image was obtained.By combination of different strategies to construct composite kernels,the spatial information was introduced into the classifier to implement the classification with the SVM and based on both the spectral information and spatial information.Hyperspectral image experiments indicate that the overall accuracy and Kappa coefficients of the proposed approach increase about 2% without increasing the training time obviously.Compared with the classifiers only using the spatial or spectral information,the proposed method shows a lot advantages.

The frequency-modulated continuous-wave (FMCW) synthetic aperture radar (SAR) is a light-weight, cost-effective, high-resolution imaging radar, which is suitable for a small flight platform. The signal model is derived for FMCW SAR used in unmanned aerial vehicles (UAV) reconnaissance and remote sensing. An appropriate algorithm is proposed. The algorithm performs the range cell migration correction (RCMC) for continuous non-chirped raw data using the energy invariance of the scaling of a signal in the scale domain. The azimuth processing is based on step transform without geometric resampling operation. The complete derivation of the algorithm is presented. The algorithm performance is shown by simulation results.

A marginalized particle filtering (MPF) approach is proposed for target tracking under the background of passive measurement. Essentially, the MPF is a combination of particle filtering technique and Kalman filter. By making full use of marginalization, the distributions of the tractable linear part of the total state variables are updated analytically using Kalman filter, and only the lower-dimensional nonlinear state variable needs to be dealt with using particle filter. Simulation studies are performed on an illustrative example, and the results show that the MPF method leads to a significant reduction of the tracking errors when compared with the direct particle implementation. Real data test results also validate the effectiveness of the presented method.

As the inevitable noises exist in actual application, the common method (MTMM) gets unacceptable result of singularity detection. In order to get more accurate singularity detection, wavelet transform shrinkage and spatially selective noise filtration methods are respectively utilized to denoise the corrupted signals. Then, the wavelet transform are applied to the two independent preprocessing signals, following that the modulus maxima are extracted for them. Based on the differences of modulus maxima dominated by noise and true signal, modulus maxima lines are picked up for the two disrelated sources. Meanwhile a proper fused and weighted manner is adapted to obtain reliable modulus maxima lines, which are directly corresponding to numbers and positions of singular points. Finally, several simulation experiments validate that the proposed algorithm obtains acceptable result of singularity detections for noisy signal, and achieves better performance over other two methods in noisy condition.

The jamming signal released by the enemy's aircraft can be explored by detecting the mutation of the received signal for radar seeker. The approximate entropy method, measuring the signal complexity, can be applied to process one frame data and to judge whether the jamming exists or not. Meanwhile, the moving-cut data approximate entropy is adopted to obtain the mutation point precisely. For three types of noise jamming and one deception jamming, the received signals of seeker radar are well analyzed respectively in intra-frame and inter-frame. The experimental results validate the effectiveness of the proposed algorithm in radar jamming detection, and for comparison, the wavelet transform is applied to detect the deception jamming, while the result validates that it is noneffective in noise condition.

When target or its parts exist micro-motion, ISAR is severely contaminated by the micro-Doppler effect caused by the micro-motion. To solve the problem, the micro-motion models of translation, rotation and vibration are analyzed, and then the time-frequency distribution of these micro-motions are discussed. After echoes are decomposed with Chirplet transformation, an algorithm, which separates signals using micro Doppler and micro Doppler rate in Chirplet parameter domain, is investigated. The algorithm is verified by simulation data of point-scatterer and real data of an airliner, nice ISAR images are obtained.

A new in time passive localization system based on multi base-line phase comparison receivers is proposed. The new system uses short base-line to avoid the long base-line phase illegibility then uses the phase differences without illegibility to get the signal arriving angle and to detect the emitter's range. The GDOP and Simulation performance have also been discussed in this paper.

An algorithm embedding bit series in the image spatial domain is studied. The algorithm determines the embedding strength using characteristics of the human visual system, so the resulting image quality is guaranteed. The algorithm utilizes the fact that natural image data is locally correlated, and uses a precede method to reduce the original image's energy, and at the same preserves the watermark information. The algorithm is discussed in detail in theory, and experimental results show consistency with theory.

The feasibility of digital Holter recording using neural networks is examined. The networks are used for data compression operations with a ratio of 1:40. Electrocardiographic data are processed by means of an artificial neural network. This necessitates a search for appropriate algorithms under conditions of decreased signal distortion. The authors describe the ideal backpropagation algorithm and procedures for the modification. >

A method of image classification using a neural network, which is a three-layer perceptron, is described. The image features are obtained by using a Zernike moment. The following conclusions are drawn from a large number of experiments about the 26 alphabets: The accuracy of the classification of the neural network is high. For various images without noise, the classical accuracy can reach 100%; for images with noises, the effect is also very good. When the ratio of the signal to noise is down to 12 dB, the classification accuracy is still 95%, with greater decrease of the ratio, the classification declines quickly. The neural network does not require many trained samples. For conditions of fewer learned samples, the effect of detection is good. >

Artificial neural network models have previously been studied in the hope of achieving human-like performance in the field of information processing. The optimized learning rule, based on the Kohonen self-organizing feature map, is modified in order to decrease the fuzziness on the edges of the topological neighbors. The authors describe the mathematical mechanisms of multidimensional detection, and its application in a radar system. High-accuracy performance is achieved, and detection is nonparametric because of the self-organizing learning process. >

This paper introduces two new types of direction-finding algorithms. They effectively combine time and spatial processing, through using the theorem of spectrum correlation. The algorithms have signal selection and overcome the shortcomings of general spatial spectral evaluation algorithms. They improve the properties of direction-finding spectral evaluation. This paper develops known spectral-correlation features for the desired signal. In cases of multipath and arbitrary numbers, arbitrary disturbances can elevate the direction of arrival of the desired signal. Computer simulations are shown as illustration.

The non-fully developed full-range sea spectrum (NDFSS) is reconstructed, and a fractal sea model based on NDFSS (NDFFM) is presented, witch is well applicable to the real sea surface. The two-scale method is used to calculate the backward radar cross section (BRCS) of the NDFFM sea surface, and the relationship between the model parameters and the BRCS is discussed in detail. Compared with the conventional fractal model and the real radar data, this model is proved to be accurate and efficient.