文献类型：会议论文

英文题名：Short-time and Spectrum Features of Noises Made by Vehicles for Recognition

作者：Ding Kai Zhang Shigong Zhang Kesheng Lei Zhen

第一作者：Ding Kai

通信作者：Ding, K[1]

机构：[1]Sci & Technol Near Surface Detect Lab, Wuxi, Peoples R China;[2]Guizhou Inst Technol, Guiyang, Peoples R China

第一机构：Sci & Technol Near Surface Detect Lab, Wuxi, Peoples R China

通信机构：corresponding author), Sci & Technol Near Surface Detect Lab, Wuxi, Peoples R China.

会议论文集：3rd International Conference on Automation Electronics and Electrical Engineering

会议日期：NOV 20-22, 2020

会议地点：Shenyang, PEOPLES R CHINA

语种：英文

外文关键词：Recognition; Vehicle types; Short-time; Spectrum

年份：2020

摘要：The recognition and classification of moving maneuvering targets play an important role in intelligent vehicle highway system. In this paper, based on short-time and spectrum technologies, some general used features of noises made by vehicles are analyzed. Short-time zero-crossing rate and short-time energy do not show obvious identifiable characteristics for recognition. While, spectrum features of noises made by different vehicles represent obvious characteristics. These features include amplitude frequency spectrum, power spectrum density, and short time amplitude frequency spectrum, Engine speed and vehicle weight can be estimated from the spectrum features. Besides, nonlinear idle engine noise features can also be used in vehicle recognition. With a good theory basis, the voice-print features, such as pitch and formant frequencies, are considered the best methods for recognition, but the two features do not really display the corresponding role. Feature extraction based on spectrum features can provide a powerful theoretical and experimental basis for the next pattern recognition.