基于仿生模式識別的DOA估計方法
A DOA Estimation Method Based on Biomimetic Pattern Recognition
-
摘要: 該文就仿生模式識別(拓撲模式識別)在非感性抽象對象的信息處理方面的應用作了一些探索，提出了一種基于仿生模式識別的DOA估計方法。這種方法的建模過程是用在實際環(huán)境下采集的訓練樣本構造人工神經網絡模型，對環(huán)境的適應能力較強，且這種方法的計算量較小，可以實現(xiàn)系統(tǒng)實時處理。實驗結果表明：在信噪比為20 dB和0 dB時，該方法的正確估計率可達100%；在信噪比降為-20 dB時，該方法仍有83%的可識別率。Abstract: In this paper, the applicability of biomimetic pattern recognition to informa-tion processing of abstract objects is studied, and then a DOA estimation method based on biomimetic pattern recognition is advanced. In this method, the output signals of an-tenna array are collected in practical conditions and expressed as feature vectors. These feature vectors are studied with the method of high dimensional geometry and the principle of biomimetic pattern recognition. By using the feature vectors as training samples ANN models are constructed. In our experiments, when SNE, is 20dB or OdB, the correct estima-tion rate is 100%; when SNR is -20dB, the correct estimation rate is 83%. The experimental results show that the proposed method has the great advantage of preferably robust and fast computation.
-
Fisher R A. Contributions to Mathematical Statistics[M]. New York: J. Wiley, 1952: 1-3. ［2］陳季鎬(美)著,邱炳章,邱華譯.統(tǒng)計模式識別[M].北京:北京郵電學院出版社,1989:1-3. ［3］Vapnik V N,Chervonenbs A Ja.Theory of Pattern Recognition[M].Moscow:Nauka(俄文),1974:1-3. ［4］Vapnik V N著,張學工譯.統(tǒng)計學習理論的本質[M].北京:清華大學出版社,2000:1-10. ［5］王守黨等.人工神經網絡的多維空間幾何分析及其理論[J].電子學報,2002,30(1):1-4. ［6］王守覺.仿生模式識別(拓撲模式識別)--一種模式識別新模型的理論與應用[J].電子學報,2002,30(10):1417-1420. ［7］王守覺等.通用神經網絡硬件中神經元基本數(shù)學模型的討論[J]..電子學報,2001,29(5):577-580. ［8］王守覺等.基于仿生模式識別的多鏡頭人臉身份確認系統(tǒng)研究[J].電子學報,2003,31(1):1-5. -
計量
- 文章訪問數(shù): 2403
- HTML全文瀏覽量: 146
- PDF下載量: 739
- 被引次數(shù): 0
下載:
