Search

Automatic ship detection in SAR imagery has been playing a significant role in the field of marine monitoring but great challenges still exist in real-time application. Despite the exciting progresses made by deep-learning techniques, most detectors failed to yield locations of fairly high quality. Moreover, the ships with variant sizes and aspects are easily omitted especially for small objects under complicated background. To alleviate the above problem, the authors propose an elaborately designed single shot detection framework combined with attention mechanism, which roughly locates the regions of interest via an automatically learned attentional map. This lay the foundation of accurate positioning of extremely small objects since the background interference can be effectively suppressed. Furthermore, a multi-level feature fusion module integrated in top-down and bottom-up manner is adopted to adequately aggregate features from not only adjacent but also distant layers. This strengthens local details and merge strong semantic information, enabling the generation of higher qualified anchors for the efficient detection of multi-scale and multi-orientated objects. Experiments on SAR ship dataset have achieved a promising result, surpassing current state-of-the-art methods.

It is well known that the application of radar is becoming more and more popular with the development of the signal technology progress. This paper lists the current radar signal research, the technical progress achieved, and the existing limitations. According to radar signal respective characteristics, the design and classification of the radar signal are introduced to reflect signal’s differences and advantages. The multidisciplinary processing technology of the radar signal is classified and compared in details referring to adaptive radar signal process, pulse signal management, digital filtering signal mode, and Doppler method. The transmission process of radar signal is summarized, including the transmission steps of radar signal, the factors affecting radar signal transmission, and radar information screening. The design method of radar signal and the corresponding signal characteristics are compared in terms of performance improvement. Radar signal classification method and related influencing factors are also contrasted and narrated. Radar signal processing technology is described in detail including multidisciplinary technology synthesis. Adaptive radar signal process, pulse compression management, and digital filtering Doppler method are very effective technical means, which has its own unique advantages. At last, the future research trends and challenges of technologies of the radar signals are proposed. The conclusions obtained are beneficial to promote the further promotion applications both in theory and practice. The study work of this paper will be useful for choosing more reasonable radar signal processing technology methods.

A new approach is developed for synthetic aperture radar (SAR) automatic target recognition based on convolutional neural network (CNN). In order to improve the low efficiency caused by fixed learning rate in CNN, the AdaDelta method is introduced to adjust the learning rate adaptively. The experimental results on Moving and Stationary Target Acquisition and Recognition (MSTAR) data sets show that convergence is significantly faster in proposed method than CNN with fixed learning rate.

Full-polarization high range resolution profile (FPHRRP) contains rich information regarding the signatures of a target which can be extracted for target recognition. The four polarization combinations data are correlated for they characterize the same target from a specific aspect angle using different polarization modes. To efficiently utilize the information for recognition performance enhancement, a joint sparse representation (JSR) based method for FPHRRP recognition is presented. Each single-polarization HRRP is represented by the atoms which are adaptively selected from its corresponding dictionary, and these atoms derived from different dictionaries have the same index set. Compared to conventional methods, the proposed method has the significant advantage of exploiting the correlation among single-polarization HRRPs to boost recognition performance. The experiment results show that the proposed method achieves promising recognition accuracy and is robust with respect to noisy observations.

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.