CAJ | 학술논문

本文提出一种基于球谐域最小方差无失真响应（MVDR）波束形成技术的三维空间远场多声源定位算法，该算法将波束形成技术推广到球谐域中，得到球谐域波束形成器的一般表达式，进而利用球谐域MVDR波束形成技术实现三维空间多声源的定位。将波束形成技术推广到球谐域中可有效地减少运算量。利用98元三线交点球麦克风阵列进行声源定位对比实验，结果表明：基于球谐域MVDR波束形成技术的声源定位算法比基于球谐域常规波束形成技术的声源定位算法具有更高的定位精度。
본문제출일충기우구해역최소방차무실진향응（MVDR）파속형성기술적삼유공간원장다성원정위산법，해산법장파속형성기술추엄도구해역중，득도구해역파속형성기적일반표체식，진이이용구해역MVDR파속형성기술실현삼유공간다성원적정위。장파속형성기술추엄도구해역중가유효지감소운산량。이용98원삼선교점구맥극풍진렬진행성원정위대비실험，결과표명：기우구해역MVDR파속형성기술적성원정위산법비기우구해역상규파속형성기술적성원정위산법구유경고적정위정도。
We propose a multiple far-field acoustic sources localization algorithm based on minimum-variance distortionless response (MVDR) beamforming in spherical harmonic domain in this paper. The beamforming technique is extended into spherical harmonic domain to generate a general expression for beamformer, and then multiple acoustic sources can be localized using MVDR beamforming technique in spherical harmonic domain. The computational complexity of the algorithm can be greatly reduced in spherical harmonic domain. Experiments are carried out using 98-element trijunctional spherical microphone array. And the results show that the proposed MVDR beamforming based algorithm has better localization performance than those regular beamforming based ones.