Chen Xiaolong, Guan Jian, He You

Transcription

1 Chen Xiaolong, Guan Jian, He You Marine Target Detection Research Group, Naval Aeronautical and Astronautical University, Yantai

2 Detection And Estimation of Low-observable Marine Target with Micromotion In Representative Domain 1. Introduction Demands in Marine Target Detection Difficulties in Marine Target Detection Micro-Doppler theory 2. Signal Model of Micromotion Target at Sea 3. Marine Target Detection via M-D Signatures High-resolution time-frequency analysis (STFRFT) Long-time coherent integration (LTCI) Sparse signal analysis (MCA) 4. Experimental Results 5. Conclusions Marine Target Detection Research Group, NAAU

3 1.1 Demands in Marine Target Detection Robust detection of weak target especially low attitude, slow, and small target in sea clutter is always a challenging subject in the field of radar signal processing, which is important for military and civil use. Different from other environment, the marine environment is rather complex. It covers all dimension space, including underwater, sea surface, low attitude, sky and outer space, and it has close relationship with the electromagnetic space. Marine Target Detection Research Group, NAAU

4 1.2 Difficulties of Marine Target Detection 1)Complex mechanism of dynamic ocean topography The marine environment is influenced by many factors such as atmosphere, hydrology, etc. The sea state changes frequently. The marine environment has an significant influence on the attenuation and propagation of electromagnetic waves, especially for radar signal. Marine Target Detection Research Group, NAAU

5 1.2 Difficulties of Marine Target Detection 3)Strong sea clutter sea spikes Sea spikes, a term to describe the strong and rapidly fluctuating events, occur with high frequency, high resolution, low grazing angles, and horizontal polarization. Sea spikes usually show nonhomogeneous, nonstationary, and time-varying properties, which would degrade the detection performance of radar detectors. Time domain Sea spikes Frequency domain 幅幅 Doppler spread 海海海海海海 ( 无海海海 ) 频频 (Hz) 海海海 ( 无海海海 ) 海海海 Sea spikes Rayleigh Lognormal Weibull K-distribution 幅幅 海海幅幅尖尖 概频 Sea spikes t (s) Sea spikes identification 幅幅 Non-Guassian property Nonstationary property Marine Target Detection Research Group, NAAU

6 1.2 Difficulties of Marine Target Detection 4)Low-observable marine target with low SCR Small size Stealth Far-range High-speed or highly mobile + Lowobservable target Amplitude Normalized Complex sea environment 浮目目海浮浮 Target!Effective solution: Sea clutter suppression; Accumulate target s energy to improve SCR Samples Range Weak signal Marine Target Detection Research Group, NAAU

7 1.2 Difficulties of Marine Target Detection 5)Difficult to Detect Moving target in sea clutter Under high oceanic conditions, or due to the pushing and control effects caused by propeller, engine, and rudder, the attitude of target may vary with the fluctuation of sea surface, which induces the effect of power modulation on radar echo. The complex motion of a rigid marine target consists of rotations and translations, that is, roll, pitch, and yaw movements. The Doppler exhibits time-varying and nonstationary properties, which is periodically frequency modulated. Target Sea clutter Time-varying Doppler frequency Marine Target Detection Research Group, NAAU

8 1.3 Micro-Doppler (m-d) theory M-D signature can be regarded as a unique signature of the target and the m-d features reflect the unique dynamic and structural characteristics. The micromotion of a subject induces Doppler frequency modulations around the carrier frequency of the reflected sensor signals. Using m-d techniques can produce identifying signatures for vehicles, machinery, animals, and human activities, which are helpful for detection and recognition. Marine Target Detection Research Group, NAAU

9 1.3 Micro-Doppler (m-d) theory It is until now that people begin to realize that the sea surface targets may also exhibit micromotions especially for high sea state and high mobility. It is found that the m-d signatures usually result from the nonuniform motion, vibrations or rotations of the marine target. However, it is rather difficult to extract and strengthen the time-varying m-d signals and Sea clutter affected by the strong sea clutter. Chen Xiaolong, et al. Applications and Prospect of Micro-motion Theory in the Detection of Sea Surface Target. Journal of Radars, 2013, 2(1): Chen Rigid Xiaolong, inflatable et al. boat Effective coherent Double-bladed integration paddle method for marine target with micromotion via phase Marine differentiation Target Detection and radon-lv's Research distribution Group, NAAU IET RSN (Special Issue: Micro-Doppler), 2015, 9(9):

10 2.1 General observation model Marine Target Detection Research Group, NAAU

11 2.1 General observation model Marine Target Detection Research Group, NAAU

12 3.1 High-resolution time-frequency analysis (STFRFT) Problem:? Fourier transform cannot reflect and extract the signature of time-varying frequency. (High resolution of time and frequency changes) 1 Frequency (Hz) M-D signal Sea clutter STFT outputs M-D signal (The #1 energy cannot be well accumulated) Time (s) Normalized frequency Solution: Using short-time fractional Fourier transform (STFRFT) to obtain high resolution spectrum of a m-d signal. Chen Xiaolong, et al. Detection and extraction of target with micro-motion in spiky sea clutter via short-time Marine fractional Target Fourier Detection transform. Research IEEE Group, Transactions NAAUon Geoscience and Remote Sensing, , (2):

13 3.1 High-resolution time-frequency analysis (STFRFT) Fractional Fourier transform (FRFT): FRFT is a powerful tool for nonstationary signal analysis, especially for LFM signal. When α=π/2, FRFT turns to Fourier transform. STFT However, FRFT contains no time information. By multiplying the signal with a window before taking the FRFT, the short-time FRFT (STFRFT) can locate signal frequency at a particular time. STFRFT Marine Target Detection Research Group, NAAU

14 3.1 High-resolution time-frequency analysis (STFRFT) Fig. 1. X-band radar Left: STFT method Right: STFRFT method Frequency (Hz) Rotation Sea clutter Translation Rotation Sea clutter Translation Time (s) Time (s) Fig. 2. S-band radar Left: STFT method Right: STFRFT method Frequency (Hz) Sea clutter 海海海海海海 and noise Moving 运目目 target M-D signal 距距 Range bin Time (s) Marine Target Detection Research Group, NAAU

15 3.2 Long-time coherent integration (LTCI) Problem:?The SCR of a m-d signal can be improved by long-time coherent integration. (Affected by ARU and DFM effects) Marine target Solution: Long-time coherent integration to simultaneously compensate the ARU and DFM effects. Chen Xiaolong, et al. Radon-fractional ambiguity function-based detection method of low-observable maneuvering target. IEEE Transactions on Aerospace and Electronic systems, 2015, 51(2). Chen Xiaolong, et al. Radon-linear canonical ambiguity function-based detection and estimation method for Marine marine Target target Detection with micromotion. Research IEEE Group, Transactions NAAU on Geoscience and Remote Sensing, , 53(4):

16 3.2 Long-time coherent integration (LTCI) Marine target Moving target detection (MTD): 长时间相参积累检测技术!Only for the uniformly moving target,limited integration gain Radon-Fourier transform (RFT) :!Invalid for the target with nonuniform or highly mobile motion. Marine Target Detection Research Group, NAAU

17 3.2 Long-time coherent integration (LTCI) For m-d modeled as LFM signal: Radon-fractional Fourier transform (RFRFT) Radon-Linear Canonical Transform (RLCT) Scale processing-rft (SPRFT) Slow-time Radar For m-d modeled as QFM signal: Radon-fractional ambiguity function (RFRAF) Radon-Linear Canonical ambiguity function (RLCAF) High-order motion or rotation Acceleration Range bins Uniform motion Marine Target Detection Research Group, NAAU

18 3.2 Long-time coherent integration (LTCI) RFRFT: RLCT: RFRAF: RLCAF: Long-time Instantaneous Auto Correlation Function Marine Target Detection Research Group, NAAU

19 3.2 Long-time coherent integration (LTCI) Relationships of RFRFT FRFT RFT and MTD: MTD: S = s ( tt, )exp( j2π ft )dt α=π/2 FRFT: S (, ) (, )d RFRFT spc ttm Kα tm u tm MTD PC m d m m = One rangebin RFT: RFRFT: [ ] S = s 2( r vt )/ ct, exp( j2π ft )dt α=π/2 RFT PC 0 0 m m d m m S s r vt at ct K t u t 2 RFRFT = PC 2( 0 0 m s m /2)/, m α( m, )d m Time Marine Target Detection Research Group, NAAU Range

20 3.2 Long-time coherent integration (LTCI) Relationships of RFRAF FRAF RAF and AF : AF α=π/2 FRAF One rangebin One rangebin RAF α=π/2 RFRAF t m Target Radar T FRAF T AF Accelerated motion Time r 0 r s (t m ) High-order motion T RFRAF T RAF T ( ) :Integration time Range bin Marine Target Detection Research Group, NAAU

21 3.2 Long-time coherent integration (LTCI) Applications in radar signal processing: Wideband radar (High range resolution) Ability to suppress clutter and noise clutter RFRFT RLCT RFRAF RLCAF SPRFT Low observable target detection (Far range stealth target) 2 3 High-speed or highly mobile target Digital phases array radar Chen Xiaolong, et al. An effective coherent integration method for marine target with micromotion via Marine PD-RLVD, Target IET Detection Radar sonar Research andgroup, Navigation, NAAUspecial issue for micro-doppler,

22 3.2 Long-time coherent integration (LTCI) For more information about the applications of LTCI in marine target detection, please refer to the following references: 1 Chen Xiaolong, et al. Maneuvering target detection via Radon-fractional Fourier transformbased long-time coherent integration. IEEE TSP, 2014, 62(4): Chen Xiaolong, et al. Radon-linear canonical ambiguity function-based detection and estimation method for marine target with micromotion. IEEE TGRS, 2015, 53(4): Chen Xiaolong, et al. Detection of a low observable sea-surface target with micromotion via the Radon-linear canonical transform. IEEE GRSL, 2014, 11(7): Chen Xiaolong, et al. Radon-fractional ambiguity function-based detection method of lowobservable maneuvering target. IEEE TAES, 2015, 51(2): Chen Xiaolong, et al. Sea clutter suppression and micromotion marine target detection via Radon-linear canonical ambiguity function. IET RSN, 2015, 9(6): Chen Xiaolong, et al. Effective coherent integration method for marine target with micromotion via phase differentiation and radon-lv's distribution IET RSN (Special Issue: Micro-Doppler), 2015, 9(9): Chen Xiaolong, et al. Radar maneuvering target detection based on scaling processing and Radon-Fourier transform. IET International radar conference, Hangzhou China, October Marine Target Detection Research Group, NAAU

23 3.3 Sparse signal analysis (MCA) Problem:? The TDF method would cost large computation burdens.? Detection and estimation performances are affected by the limited time-frequency resolution of TFD method Frequency (Hz) Sea 海海海 clutter Time (s) Chirp (Hz/s) 调频频 (Hz/s) 加加运 M-D 运动动 signal M-D 转运动动 signal 海海海 频频 (Hz) Frequency (Hz) Solution: Distinguish sea clutter and m-d signal by their different sparse properties. (Morphological component analysis, MCA) Chen Xiaolong, et al. Application of the sparse decomposition to micro-motion target detection embedded in sea clutter, 2013 International Radar Conference, Australia, September 2013, Chen Xiaolong, et al. Detection of Marine Target with Quadratic Modulated Frequency Micromotion Signature via Marine Morphological Target Detection Component Research Analysis, Group, CoSeRa, NAAU Italy,

24 3.3 Sparse signal analysis (MCA) Fig. 1. Description of the sea clutter dataset. (a) Range-Doppler analysis (b) Time-frequency analysis of target Fig. 2. Detection results comparison via MCA with different dictionaries (P fa =10-4 ). (a) FT dictionary (b) Chirp dictionary (c) PD+chirp dictionary Marine Target Detection Research Group, NAAU

25 Marine Target Detection Research Group The main research field includes: Sea clutter collection, analysis, modeling and suppression; Marine target characteristics analysis, modeling and feature extraction; Marine targets detection, tracking and recognition; New mechanism and new methods research for marine radar detection; Sea trial for radar detection and evaluation. Marine Target Detection Research Group, NAAU

26 Marine target detection trial field Marine Target Detection Research Group, NAAU

27 4. Experimental Results MTD method MTD outputs v0 (knots)=9.819 Sea clutter and noise Marine target Freqency(Hz) SPRFT method Tn=3s, Pulses number=1500 SCR out =12.75dB, Computing time=56.7ms SPRFT outputs Marine target Threshold Tn=0.5s, Pulses number=250 SCR out =1.25dB, Computing time=10ms Suppressed clutter and noise Marine target Threshold Frequency (Hz) RFT method RFT outputs 0.6 Sea clutter and noise Frequency (Hz) Threshold Tn=3s, Pulses number=1500 SCR out =5.83dB, Computing time=43.4ms Marine Target Detection Research Group, NAAU

28 4. Experimental Results 2)X-band CSIR datasets Sea 海海海 clutter Sea clutter Radar 目目目海 returns GPS GPS 距距 Target 时时 (s) Radar deployment Range versus time Time-frequency analysis WaveRider Radar RIB returns 目目目海 4 Target Sea 海海 clutter 海 GPS GPS 运运运运 Sea clutter Marine Target Detection Research Group, NAAU

29 5. Conclusions The micro-doppler signatures of marine target can provide extra Doppler and motion information, which are very useful for target detection and classification. It is difficult to extract and strengthen the micro-doppler signatures due to the nonstationary and time-varying properties and heavy sea clutter. The STFRFT, long-time coherent integration and sparse signal analysis technique can be employed to improve SCR and obtain high resolution of time-frequency. More work can be done considering radar system and configurations. Marine Target Detection Research Group, NAAU

30 Looking forward to your cooperation and guidance! Chen Xiaolong Marine Target Detection Research Group, Naval Aeronautical and Astronautical University