Scholarlink

<p>Accurate identification of the origin of salmon is of great importance for food safety and market regulation; however, traditional testing methods are generally characterised by cumbersome procedures, high costs, and a destructive nature. Consequently, this study established a low-cost, rapid, and non-destructive testing method. Spectral imaging technology was employed to analyse salmon samples from four different origins, and a snapshot spectral imaging system was used to capture both spectral data and image texture information. Feature wavelengths were screened using the Successive Projections Algorithm (SPA), while texture features were extracted based on the Grey-level Co-occurrence Matrix (GLCM). Multi-source feature fusion was achieved through vector normalisation, and a least squares support vector machine (LS-SVM) classification model was constructed. The experimental results indicated that the classification accuracy of the model based on spectral imaging data reached 99.52%. While achieving high-precision detection, the method offered significant cost-effectiveness, providing an effective and viable approach for the non-destructive identification of salmon origin.</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

Identifying the origin of salmon and trout using spectroscopic and spectral imaging techniques

  • 期刊论文

  • 发表日期：

    2026年6月17日

查看全文
  • Jiyang Li,
  • Yu Ren,
  • Hengyuan Zhang,
  • Yangyang Kong,
  • Hongxing Cai

摘要

Accurate identification of the origin of salmon is of great importance for food safety and market regulation; however, traditional testing methods are generally characterised by cumbersome procedures, high costs, and a destructive nature. Consequently, this study established a low-cost, rapid, and non-destructive testing method. Spectral imaging technology was employed to analyse salmon samples from four different origins, and a snapshot spectral imaging system was used to capture both spectral data and image texture information. Feature wavelengths were screened using the Successive Projections Algorithm (SPA), while texture features were extracted based on the Grey-level Co-occurrence Matrix (GLCM). Multi-source feature fusion was achieved through vector normalisation, and a least squares support vector machine (LS-SVM) classification model was constructed. The experimental results indicated that the classification accuracy of the model based on spectral imaging data reached 99.52%. While achieving high-precision detection, the method offered significant cost-effectiveness, providing an effective and viable approach for the non-destructive identification of salmon origin.