Page 101 - 《应用声学》2020年第1期

第 39 卷 第 1 期 Vol. 39, No. 1
2020 年 1 月 Journal of Applied Acoustics January, 2020

⋄ 研究报告 ⋄

基于子带信号瞬时频率的特征提取及其在

车型分类中的应用 ∗

陈建新 1† 尹雪飞 1 陈克安 2

(1 西北工业大学电子信息学院 西安 710129)
(2 西北工业大学航海学院 西安 710072)

摘要：车型识别是智能交通系统的关键技术之一，具有重要应用价值。针对车辆噪声信号的复杂性，提出了一
种基于相位信息和能量信息融合的车型分类方法。通过耳蜗滤波器组将车辆噪声信号分解成窄带信号，为了
避免相位卷绕问题，利用傅里叶变换性质结合相位一阶导数估计窄带信号的瞬时频率并提取瞬时频率特征。
该特征能够有效地完成车型分类，通过将瞬时频率特征和对数能量联合，进一步提高了分类准确率。
关键词：车型分类；耳蜗滤波器组；瞬时频率；对数能量
中图法分类号: TN911.72 文献标识码: A 文章编号: 1000-310X(2020)01-0097-07
DOI: 10.11684/j.issn.1000-310X.2020.01.012

Feature extraction based on instantaneous frequency of subband signal and its
application in vehicle classification

CHEN Jianxin 1 YIN Xuefei 1 CHEN Ke’an 2

(1 School of Electronics and Information, Northwestern Polytechnical University, Xi’an 710129, China)
(2 School of Marine Technology, Northwestern Polytechnical University, Xi’an 710072, China)
Abstract: Vehicle recognition is one of the key technologies of intelligent transportation system and has im-
portant application value. Aiming at the complexity of vehicle noise signal, a vehicle classification method
based on phase information and energy information fusion was proposed. The vehicle noise signal was decom-
posed into narrow-band signals by cochlear filter banks. To avoid phase wrapping problem, the instantaneous
frequency of narrow-band signals was estimated by combining the Fourier transform property with the first
derivative of phase, and the instantaneous frequency (IF) feature was extracted. This feature can effectively
complete vehicle classification. By combining IF feature with logarithmic energy, the classification accuracy is
further improved.
Keywords: Vehicle classification; Cochlear filter bank; Instantaneous frequency; Logarithmic energy

2019-04-17 收稿; 2019-08-15 定稿
国家自然科学基金项目 (11574249)
∗
作者简介: 陈建新 (1993– ), 男, 河南人, 硕士研究生, 研究方向: 信号与信息处理。
† 通信作者 E-mail: 2016201348@mail.nwpu.edu.cn