Rui Min

Application of AdaBoost in polarimetric SAR image classification

In this paper, a method of polarimetric SAR image classification based on polarimetric decomposition and AdaBoost algorithm is proposed. The proposed method improves classification accuracy and speed. AdaBoost algorithm, as a robust learner with high accuracy, can fully utilize the polarimetric features to achieve image classification. In simulated tests, the proposed method is observed to produce improved classification accuracy and speed, compared with H /α̅ classification algorithm.

SAR target recognition via incremental nonnegative matrix factorization with L p sparse constraint

In this paper a novel incremental nonnegative matrix factorization (INMF) for SAR target recognition is proposed in order to overcome the defect that conventional methods have in online processing. When training samples increase, unlike conventional NMF-based methods computing both original and new samples to retrain a new model, INMF just computes the new samples to update the trained model incrementally, which can avoid repetitive learning original samples and reduce the computation cost. Meanwhile, the INMF with L p constraint is proposed by setting the updating process under L p sparse constraint for matrices decomposition. The proposed L p -INMF can solve the problem of the computational cost increasing along with the sample increasing which is common in traditional methods. Experiment results on MSTAR data verify that the recognition performance obtained by L p -INMF outperforms other traditional methods, and the recognition efficiency can be improved.

Radar maneuvering target detection based on two steps scaling and fractional Fourier transform

Abstract This paper considers the detection problem of radar maneuvering target with the range migration (RM) and Doppler frequency migration (DFM) and proposes an coherent integration method based on two steps scaling and fractional Fourier transform, i.e., TSS-FrFT. To be specific, this method corrects the quadratic RM (QRM) and DFM based on FrFT and a scaling processing of the rotation angle firstly. Then, it eliminates the coupling between the range frequency variable and FrFT domain (FrFD) variable via a second scaling processing. In the end, the coherent integration is achieved by the inverse Fourier transform (IFT) with respect to the range frequency variable. Compared with the existing methods, the proposed TSS-FrFT method has superior detection performance and low computational complexity. Moreover, considering the sudden change of the target acceleration within the coherent integration interval, this paper also presents an acceleration matched filter (AMF) to remove the Doppler frequency rate change (DFRC) caused by this acceleration change. Computer simulation results are also given to demonstrate the effectiveness of the proposed method from different aspects, i.e., computational complexity, coherent integration for a single target and multiple targets and detection performance for a uniform acceleration and non-uniform accelerations.

Micro Drone Detection and Parameters Estimation Based on Micro-Doppler of Blades

Small unmanned aerial vehicles (UAVs) or micro drones are widely used for many applications in these years. But the misuse of micro drones may cause security issues. The problem of micro drone detection and parameters estimation is considered in this paper. Micro Doppler signatures of the drone’s rotating rotor blades are applied to identify micro drone. Firstly, the signal mathematical model of drone’s rotating multi-rotor blades is built and the flashes are used for detection. Then, the relationship between the micro Doppler signatures and the parameters of drones is analyzed. Furthermore, parameters, such as the number of blades, the number of rotor, the rotation rate and the length of blade, are estimated in this article. Finally, the results show that the effectiveness of the proposed method.

Geospatial Object Partition Based on Angular Second Moment Kernel

Improvements in Synthetic Aperture Radar (SAR) image collection has revealed the ability to semantically describe scene complexity and abundant details. It is difficult for the traditional pixel-based methods to partition geospatial objects. A practical geospatial object partition method based on angular second moment kernel (ASMK) for SAR image is proposed. Firstly, a new kernel termed ASMK is designed in order to obtain accurate classification results. Then, based on classification results, river and urban areas as typical geospatial objects are partitioned. In order to obtain urban border accurately, a likelihood function to evaluate the possibility that one pixel belongs to urban area is established. Results of experiments with high-resolution TerraSAR-X spotlight data of the urban of Rosenheim in Germany demonstrate that the proposed method’s effectivity and accuracy in object partition.

Polarimetric-spatial classification of PolSAR images based on composite kernel feature fusion

Composite kernel feature fusion is proposed in this paper for solving the classification of polarimetric synthetic aperture radar (PolSAR) images problem. The main idea is that the method of composite kernel encodes diverse information within a new kernel matrix and tunes the contribution of different type of features. The proposed approach is tested on Flevoland PolSAR data set. Experimental results verify the benefits of using both of polarimetric and spatial information by composite kernel feature fusion for the classification of PolSAR images.