摘要
随着小康社会的全面建成,居民对生活水平的要求已经从温饱过渡到高质量,特别是对饮食安全问题尤为重视,但是“变质肉”、“掺假肉”、“添加肉”和“注水肉”等食品质量安全事故频发,已经严重威胁到了我国居民生命安全并阻碍了市场良性发展。目前,红肉质量检测主要依托复杂的理化实验完成,对红肉产品具有强烈的破坏性,仅适用于市场监管部门的抽查。高光谱技术作为一种原位无损、高通量、快速的智能检测技术,为解决传统检测方法在红肉生产销售全产业链中缺乏操作可行性提供了有效的技术手段,可以极大的促进我国红肉质量安全监管体系的发展与健全。综述了近几年国内外关于红肉质量高光谱无损检测研究的最新进展:首先,总结了基于高光谱无损检测技术构建红肉质量无损检测模型的优缺点,其优势是具有图谱合一、高分辨率等特性,为模型多样性提供良好的数据基础;其劣势是高光谱数据的冗余度高、信噪比低、非线性强,对模型效率造成一定影响。然后,重点分析了红肉质量无损检测建模中关键算法的研究进展:(1)感兴趣区域一般通过手动获取,感兴趣区域的自动分离方法是目前研究的重点之一;(2)光谱预处理算法主要通过观察光谱信号或根据建模效果反推选择,目前还未形成标准通用的预处理算法;(3)综合红肉光谱和图像特征,能够全面反映红肉的质量特性,为建模提供了良好的数据基础;(4)线性模型的发展应用较为成熟,稳定性较好,但是面向复杂的红肉质量检测环境,非线性模型的研究潜力更加良好。最后,通过综述近几年红肉质量的高光谱无损检测研究成果,展望了未来的研究中,提高算法自动化、充分利用图谱信息、加强非线性模型的应用将成为重点研究方向。
With the complete construction of the All-Roundly Well-off Society in China,residents have higher and higher requirements for the quality of life,especially for food safety.However,food quality and safety accidents such as“deteriorated meat”,“adulterated meat”,“added meat”and“water-injected meat”frequently occurring to threaten the life safety of Chinese residents seriously and hinder healthy development of the market.The quality test method of red meat is a physical and chemical experiment that seriously damages the samples and is only applicable to the spot check of the market supervision department.Hyperspectral technology is a kind of in-situ non-destructive,high-throughput and,fast intelligent detection technology which provides effective technology for solving the low operational feasibility of traditional detection methods.It greatly promotes the development and improvement of the quality and safety supervision system of red meat in China.This paper aims to review the research progress of hyperspectral technology in non-destructive detection of red meat quality.Firstly,the advantages and disadvantages of the red meat quality model based on the Hyperspectral technique are summarized.Its advantage is high resolution and a combination of image and spectrum,which will provide better data for the model.Then,the key algorithms in the model are analyzed:(1)Due to regions of interest obtained manually,automatic separation of regions of interest will be one of the focus of research;(2)The spectral preprocessing algorithm is mainly selected by observing the spectral signal or extrapolating by model,so there is no standard general preprocessing algorithm;(3)The combination of spectrum and image features can more comprehensively describe the quality of red meat and provide a batter basis for modeling;(4)The linear model is more mature and stable,but the research potential of nonlinear model is better for the complex environmental factors in red meat quality detection.Finally,the future development dire
作者
白雪冰
马殿坤
张梦杰
马瑞芹
BAI Xue-bing;MA Dian-kun;ZHANG Meng-jie;MA Rui-qin(College of Information and Electrical Engineering,China Agricultural University,Beijing 100083,China)
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2022年第7期1993-1998,共6页
Spectroscopy and Spectral Analysis
基金
财政部和农业农村部:国家现代农业产业技术体系项目(CARS-38)
欧盟项目(DCI:ASIE/2012/307-186)资助。
关键词
高光谱技术
红肉质量安全
特征融合
无损检测
Hyperspectral technology
The red meat quality
Feature fusion
Non-destructive testing
