Page 152 - 应用声学2019年第5期
P. 152

904                                                                                  2019 年 9 月


                                               表 2  测试结果与实际情况的对比
                               Table 2 comparison of measured data by EDM and real data

                                                                                   声源到各个墙面距离
                        4 个反射面           镜像源坐标/m             理论镜像源坐标/m
                                                                                   与实际距离的误差/m
                         左墙面           (−3.11, 2.12, 1.56)   (−3.07,.2.00, 0.82)         0.20
                         右墙面            (2.37, 1.93, 1.36)    (1.53, 2.00, 0.82)         0.28
                         天花板            (0.96, 1.81, 4.33)    (0.79, 2.00, 3.43)         0.10
                          地板            (0.96, 2.00,−1.5)    (0.79, 2.00,−0.82)          0.45
                          平均                                                             0.26

                                                                   small array sound source localization[J]. IEEE Transac-
             3 结论                                                  tions on Audio, Speech and Language Processing, 2010,
                                                                   18(7): 1781–1792.
                 本文提出了一种依托 image 模型,利用到达时                        [3] Dokmanić I, Vetterli M. Room helps: acoustic localiza-
             间差最小二乘法为一阶反射波进行分类,之后利用                                tion with finite elements[C]. IEEE International Confer-
                                                                   ence on Acoustics, Speech and Signal Processing, 2012:
             室内声场模型实现室内空间结构重构的算法。相比
                                                                   2617–2620.
             于其他室内空间建图的方法,这种方法原理简单,                              [4] Dokmanić I, Lu Y M, Vetterli M. Can one hear the shape
             利于实现,而且不需要额外的专业器材,只需要利                                of a room: the 2-D polygonal case[C]. IEEE International
             用常见的传声器组成的传声器阵列和普通全向音                                 Conference on Acoustics, Speech and Signal Processing,
                                                                   2011: 321–324.
             响就可以实现较好的效果。在此基础上,为了进一
                                                                 [5] Markovica D, Antonacci F, Sarti A, et al. Estimation of
             步提高室内建图的精度,今后可以考虑以下方案做                                room dimensions from a single impulse response[C]. IEEE
             出改进:(1) 采用全向性更好的 12 面体声源;(2) 增                        Workshop on Applications of Signal Processing to Audio
             加传声器阵元的数量以及利用更有效的三维阵型;                                and Acoustics, 2014: 1–4.
                                                                 [6] Antonacci F, Filos J, Thomas M R P, et al. Inference of
             (3) 利用TDOA和其他定位算法的结合实现更加精
                                                                   room geometry from acoustic impulse responses[J]. IEEE
             准的远场定位。本文实现了具有规则几何图形的重                                Transactions on Audio, Speech and Language Processing,
             建,考虑到后期复杂场景的重建,可以考虑引入移动                               2012, 20(10): 2683–2695.
                                                                 [7] Ba D, Ribeiro F, Zhang C, et al. L1 regularized room
             机器人测量的方法,将非直线墙(例如,折线或曲线)
                                                                   modeling with compact microphone arrays[C]. 2010 IEEE
             分为一个个小的直线,每隔一段时间发射信号并接                                International Conference on Acoustics, Speech and Signal
             收来探测绘制墙线。这将是未来实现复杂室内构图                                Processing, 2010: 157–160.
             的一个发展方向。                                            [8] Dokmanić I, Parhizkar R, Walther A, et al.  Acoustic
                                                                   echoes reveal room shape[J]. Proceedings of the National
                                                                   Academy of Sciences of the United States of America,
                                                                   2013, 110(30): 12186–12191.
                            参 考     文   献
                                                                 [9] Allen J, Berkley D. Image method for efficiently simu-
                                                                   lating small room acoustics[J]. Journal of the Acoustical
              [1] Ribeiro F, Ba D, Zhang C, et al. Turning enemies into  Society of America, 1979, 65(4): 943–950.
                 friends: using reflections to improve sound source local-  [10] Ian S. Ground based wireless positioning[M]. Piscataway,
                 ization[C]. IEEE International Conference on Multimedia  N. J: Wiley-IEEE Press, 2009: 135–136.
                 and Expo., 2010: 731–736.                      [11] Farina A. Simultaneous measurement of impulse response
              [2] Ribeiro F, Zhang C, Florencio D A, et al. Using reverber-  and distortion with a swept-sine technique[C]. 108th Au-
                 ation to improve range and elevation discrimination for  dio Engineering Society Convention, 2000: 18–22.
   147   148   149   150   151   152   153   154   155   156