Scholarlink

<p>For an in-vehicle active noise control (ANC) system, the reduction target is around the human ear. But the ear position will change when a driver alternate his sitting position, talk to the passengers, etc. These changes obviously affect the secondary path from the ear to the microphone. Thus a fixed error microphone is unable to ensure a steady noise reduction. Aiming at offering a virtual moving microphone, an active noise control system with visual tracking was proposed in this paper. Firstly, a cascade classifier is designed to recognize the human ears. An improved haar-like features and AdaBoost algorithm was utilized to decrease the training time. Then the location of the driver’s ear can be calculated referencing to the camera’s spatial coordinates. The secondary path transfer function is updated according to these changed location values. Finally, the ANC algorithm generate offset noise adapting to the updated ear position. Using the road noise collected from the real vehicle as the reference noise for noise reduction experiments. The test results show that the system can maintain a stable and consistent noise reduction within the ear moving range of 0 ~ 200&#xa0;mm. The highest 14&#xa0;dB noise reduction can be realized at the low frequency of 150&#xa0;Hz or less.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Study of an Active Noise Control System with Continuous Tracking of the Human Ear and Noise Segmentation Control

  • 期刊论文

  • 发表日期：

    2025年1月24日

查看全文
  • Hehua Su,
  • Jiang Liu,
  • Anqing Liu,
  • Baogang Li

摘要

For an in-vehicle active noise control (ANC) system, the reduction target is around the human ear. But the ear position will change when a driver alternate his sitting position, talk to the passengers, etc. These changes obviously affect the secondary path from the ear to the microphone. Thus a fixed error microphone is unable to ensure a steady noise reduction. Aiming at offering a virtual moving microphone, an active noise control system with visual tracking was proposed in this paper. Firstly, a cascade classifier is designed to recognize the human ears. An improved haar-like features and AdaBoost algorithm was utilized to decrease the training time. Then the location of the driver’s ear can be calculated referencing to the camera’s spatial coordinates. The secondary path transfer function is updated according to these changed location values. Finally, the ANC algorithm generate offset noise adapting to the updated ear position. Using the road noise collected from the real vehicle as the reference noise for noise reduction experiments. The test results show that the system can maintain a stable and consistent noise reduction within the ear moving range of 0 ~ 200 mm. The highest 14 dB noise reduction can be realized at the low frequency of 150 Hz or less.