基于音频技术的肉鸡采食量检测方法研究
作者:
作者单位:

作者简介:

通讯作者:

基金项目:

中欧政府间合作项目(SFS-46-2017);国家重点研发计划项目(2017YFD0701602-2)


Research of detection method for broiler chicken feed intake based on audio technology
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    【目的】通过研究肉鸡采食音频提出一种基于音频技术的肉鸡采食量检测方法,以摆脱目前我国肉鸡采食量数据主要是人工测量群体采食量的现状,为准确获取肉鸡采食量信息提供技术支持。【方法】录音笔采集到的采食音频经分帧加窗、端点检测等预处理后,将有效声音片段提取出来,依托不同声音的功率谱密度曲线差异,使用单分类支持向量机(OC-SVM)对提取出的有效声音片段进行分类识别。利用音频技术检测肉鸡进食时的啄食次数,分析确定啄食次数与采食量的关系,利用啄食次数与采食量的高相关性计算肉鸡采食量。【结果】利用音频技术检测的肉鸡啄食次数与采食量高度相关,决定系数(R2)=0.982 5。啄食次数计算正确率为94.58%,采食量计算正确率为91.37%。【结论】该方法可用于肉鸡采食量测定。

    Abstract:

    【Objective】 To propose a method to detect feed intake based on broiler chicken feeding audio, overcome the situation that feed intake data of broiler chickens in China nowadays were mainly population feed intake measured manually and provide a strong technical support for accurately measuring broiler chicken feed intake.【Method】 The feed intake audio collected by voice recorder was conducted framing, windowing, endpoint detection and other preprocessings, the effective sound fragments were picked up. Then one-class support vector machine (OC-SVM) was used to classify extracted sound fragments based on power spectral density curve differences of different sounds. The pecking number during feeding process was detected through audio technology. The relationship between pecking number and feed intake was explored, and the feed intake was calculated by their strong correlation.【Result】 The pecking number of broiler chicken detected through audio technology was highly correlated with feed intake, and the determination coefficient (R2) was 0.982 5. The calculation accuracy rates of pecking number and feed intake were 94.58% and 91.37% respectively.【Conclusion】 This method can be used for feed intake measurement of broiler chicken.

    参考文献
    相似文献
    引证文献
引用本文

杨稷,沈明霞,刘龙申,陆明洲,何灿隆,李嘉位.基于音频技术的肉鸡采食量检测方法研究[J].华南农业大学学报,2018,39(5):118-124

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:2018-03-25
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2018-07-18