Scholarlink

<p>Portable near-infrared spectroscopy (NIRS) was utilized, combined with chemometrics, for the rapid identification of a specific geographical variety of pork, as well as the common landrace type, just from the chunked fillets. Additionally, rapid quality grading of the pure Rongchang pork (Grade I, II, and III) was also carried on. Various preprocessing and spectral feature extraction methods were utilized and optimized to minimize redundant multivariate information. Type recognition models were established for the collected pork samples (Rongchang 40, hybrid Rongchang, and Landrace 30). Results indicated that the classification accuracy, average precision, average recall, etc., of the max-min-CARS-RF, and max-min-PCA-LPboots models for pork types identification were all 100%. For the simplicity of modeling, the max-min-CARS-RF method was considered as the optimal identification method for pork types. Furthermore, the 1D-RF model outperformed for three Rongchang pork quality grading. This study demonstrated portable NIRS sensor combined with a proper chemometrics for rapid identification of geographical pork types and quality grades on chunked meat was practical.</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

Spectra to species: recognition of geographically originated pork by portable NIRS coupled with chemometrics

  • 期刊论文

  • 发表日期：

    2025年10月11日

查看全文
  • Huijuan Cao,
  • Yanchang Bai,
  • Shengrui Zhang,
  • Heng Yu,
  • Yinghao Gao,
  • Yue Huang

摘要

Portable near-infrared spectroscopy (NIRS) was utilized, combined with chemometrics, for the rapid identification of a specific geographical variety of pork, as well as the common landrace type, just from the chunked fillets. Additionally, rapid quality grading of the pure Rongchang pork (Grade I, II, and III) was also carried on. Various preprocessing and spectral feature extraction methods were utilized and optimized to minimize redundant multivariate information. Type recognition models were established for the collected pork samples (Rongchang 40, hybrid Rongchang, and Landrace 30). Results indicated that the classification accuracy, average precision, average recall, etc., of the max-min-CARS-RF, and max-min-PCA-LPboots models for pork types identification were all 100%. For the simplicity of modeling, the max-min-CARS-RF method was considered as the optimal identification method for pork types. Furthermore, the 1D-RF model outperformed for three Rongchang pork quality grading. This study demonstrated portable NIRS sensor combined with a proper chemometrics for rapid identification of geographical pork types and quality grades on chunked meat was practical.