雷达学报
雷達學報
뢰체학보
JOURNAL OF RADARS
2014年
5期
497-504
,共8页
低分辨机载雷达%空地运动目标分类识别%分形特征%相位调制特征%支持向量机(SVM)
低分辨機載雷達%空地運動目標分類識彆%分形特徵%相位調製特徵%支持嚮量機(SVM)
저분변궤재뢰체%공지운동목표분류식별%분형특정%상위조제특정%지지향량궤(SVM)
Low-resolution airborne radar%Air/ground moving target classification%Fractal feature%Phase modulation feature%Support Vector Machine (SVM)
分类识别技术是雷达当今和未来发展的重要需求，也是雷达的关键技术之一。目前研究较多的是基于宽带信号的目标识别，对雷达系统和目标信噪比具有较高的要求，且对角度非常敏感。针对低分辨机载雷达工作在下视模式下，慢速飞行目标和地面运动目标由于具有相似的多普勒速度和雷达散射截面(RCS)，使得其对机载雷达慢速飞行目标检测、跟踪和识别形成干扰，该文提出了一种基于窄带分形和相位调制特征的机载雷达空地运动目标分类识别算法。文中以实测试飞数据进行分析验证，以支持向量机(SVM)为分类器，试验结果表明，该方法能对机载雷达直升机、汽车运动目标进行有效分类识别，当SNR≥15 dB时，平均分类识别率在89%以上。
分類識彆技術是雷達噹今和未來髮展的重要需求，也是雷達的關鍵技術之一。目前研究較多的是基于寬帶信號的目標識彆，對雷達繫統和目標信譟比具有較高的要求，且對角度非常敏感。針對低分辨機載雷達工作在下視模式下，慢速飛行目標和地麵運動目標由于具有相似的多普勒速度和雷達散射截麵(RCS)，使得其對機載雷達慢速飛行目標檢測、跟蹤和識彆形成榦擾，該文提齣瞭一種基于窄帶分形和相位調製特徵的機載雷達空地運動目標分類識彆算法。文中以實測試飛數據進行分析驗證，以支持嚮量機(SVM)為分類器，試驗結果錶明，該方法能對機載雷達直升機、汽車運動目標進行有效分類識彆，噹SNR≥15 dB時，平均分類識彆率在89%以上。
분류식별기술시뢰체당금화미래발전적중요수구，야시뢰체적관건기술지일。목전연구교다적시기우관대신호적목표식별，대뢰체계통화목표신조비구유교고적요구，차대각도비상민감。침대저분변궤재뢰체공작재하시모식하，만속비행목표화지면운동목표유우구유상사적다보륵속도화뢰체산사절면(RCS)，사득기대궤재뢰체만속비행목표검측、근종화식별형성간우，해문제출료일충기우착대분형화상위조제특정적궤재뢰체공지운동목표분류식별산법。문중이실측시비수거진행분석험증，이지지향량궤(SVM)위분류기，시험결과표명，해방법능대궤재뢰체직승궤、기차운동목표진행유효분류식별，당SNR≥15 dB시，평균분류식별솔재89%이상。
Radar Target Recognition (RTR) is one of the most important needs of modern and future airborne surveillance radars, and it is still one of the key technologies of radar. The majority of present algorithms are based on wide-band radar signal, which not only needs high performance radar system and high target Signal-to-Noise Ratio (SNR), but also is sensitive to angle between radar and target. Low-Resolution Airborne Surveillance Radar (LRASR) in downward-looking mode, slow flying aircraft and ground moving truck have similar Doppler velocity and Radar Cross Section (RCS), leading to the problem that LRASR air/ground moving targets can not be distinguished, which also disturbs detection, tracking, and classification of low altitude slow flying aircraft to solve these issues, an algorithm based on narrowband fractal feature and phase modulation feature is presented for LRASR air/ground moving targets classification. Real measured data is applied to verify the algorithm, the classification results validate the proposed method, helicopters and truck can be well classified, the average discrimination rate is more than 89% when SNR≥15 dB.
