WO2011041925A1 - 基于多传感信息融合的名优茶品质仪器智能化审评方法 - Google Patents

基于多传感信息融合的名优茶品质仪器智能化审评方法 Download PDF

Info

Publication number
WO2011041925A1
WO2011041925A1 PCT/CN2009/001538 CN2009001538W WO2011041925A1 WO 2011041925 A1 WO2011041925 A1 WO 2011041925A1 CN 2009001538 W CN2009001538 W CN 2009001538W WO 2011041925 A1 WO2011041925 A1 WO 2011041925A1
Authority
WO
WIPO (PCT)
Prior art keywords
tea
sensor
famous
quality
information
Prior art date
Application number
PCT/CN2009/001538
Other languages
English (en)
French (fr)
Inventor
赵杰文
陈全胜
蔡健荣
黄星奕
邹小波
Original Assignee
江苏大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 江苏大学 filed Critical 江苏大学
Publication of WO2011041925A1 publication Critical patent/WO2011041925A1/zh

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/02Food
    • G01N33/14Beverages
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/25Fusion techniques
    • G06F18/253Fusion techniques of extracted features
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/60Type of objects
    • G06V20/68Food, e.g. fruit or vegetables

Definitions

  • the present invention relates to an intelligent evaluation method for an instrument for the quality of a famous tea, and particularly relates to an intelligent evaluation of a famous tea quality instrument based on multi-sensor information fusion. Evaluation method.
  • BACKGROUND OF THE INVENTION Due to the high price of famous teas and the large profit margin of products, the phenomenon of shoddy and fakes in the famous tea market has been caused to some extent.
  • the famous tea products are mixed with fish, and the quality is mixed. It brings consumers a risk of purchase, which not only damages the interests of consumers, but also is not conducive to the brand maintenance of famous tea. For a long time, sensory evaluation has been regarded as the most basic method for testing the quality of tea.
  • the content of the review by trained professional reviewers generally includes color, fragrance, taste, shape and leaf bottom (tea residue). Representing different aspects of tea quality.
  • the sensory review has the following two limitations: 1 Sensory review is an expert behavior after all, and the training of experts is a long-term training and experience acquisition process. It is necessary to train a well-trained sensory review expert. At large cost, ordinary consumers generally do not have this ability; 2 The review is subject to greater interference from the objective factors, even with well-trained senior review experts, who feel the sensitivity of the organ is also affected by experience, gender, mental state, body The situation and even the regional environment and other factors change, which affects the accuracy and stability of the sensory evaluation results.
  • the application of electronic nose and electronic tongue technology in tea quality testing is still only in the laboratory stage, and there is no relevant patent literature. And in the tea quality testing, all the methods used are single detection methods, which have certain limitations.
  • the indicators that reflect the quality of tea are multi-faceted, including sensory indicators such as color, fragrance, taste, shape, etc., and a single detection method often cannot fully describe an object, only one aspect can be described. The limitations brought about will inevitably affect the accuracy and stability of the test results.
  • the invention provides an intelligent evaluation method for a famous tea quality instrument based on multi-sensor information fusion, which simulates a person by using three kinds of sensors, namely, computer vision (visual sensor), electronic nose (olfactory sensor) and electronic tongue (taste sensor).
  • Eye, nose and tongue three sensory organs Do not collect the sensor information that reflects the quality index of the famous tea; associate these feature vectors with the results of the expert sensory review, and use the appropriate learning and training to realize the sensor's identity description of the famous tea quality indicators;
  • the three kinds of information are integrated and balanced, and a multi-information fusion model for intelligent review of famous tea instruments is constructed by appropriate pattern recognition method.
  • the sample to be tested is extracted by the corresponding data acquisition and feature vector, and the extracted feature vector is substituted into the pre-established model to intelligently evaluate the quality of the unknown sample.
  • the invention can imitate the human sensory evaluation method and intelligently evaluate the quality of the famous tea.
  • the object of the present invention is to provide an intelligent evaluation of a famous tea quality instrument based on multi-sensor information fusion.
  • the evaluation method uses three kinds of sensors: computer vision, electronic nose and electronic tongue to simulate the three sensory organs of human eyes, nose and tongue, respectively, and collects various sensory information that can reflect the quality index of famous tea; respectively, these feature vectors are respectively combined with experts.
  • the results of the sensory review are related. Through appropriate learning and training, the sensor's identity description of the quality tea indicators is realized.
  • the intelligent evaluation method for the famous tea quality instrument based on multi-sensor information fusion according to the present invention is carried out according to the following steps:
  • (1) For a certain kind of famous tea first select a representative sample (the number of samples is greater than 100), and the professional will conduct a sensory evaluation on the quality indicators of the famous tea, and establish a standard based on the evaluation results. Database; wherein the establishment of the standard database, sensory evaluation of famous tea samples by more than 3 professional reviewers; sensory evaluation indicators mainly include 4 factors (dry tea color, shape, aroma and taste of tea soup, etc.); The reviewer scores the factors one by one, the average score of the professional reviewers as the final score of the quality factor, builds a standard database through the data obtained by the professional sensory review, and establishes a service for the next intelligent review model. .
  • the collected tea raw materials are evenly laid in a glass container, and then placed in a self-made light source box to start image data collection; 3 in the taste sensor data collection process, the tea raw materials for image data collection are placed in a beaker, added Deionized water, 10 ⁇ 20min in water bath at 60 ⁇ 80 °C, filter and centrifuge, take 5mL of supernatant and dilute 20 times, pipette into the beaker of electronic tongue system for data acquisition of electronic tongue; data on electronic tongue During the collection process, the taste sensor is immersed in the tea soup solution, at which time the different functional groups on the sensor biofilm and the odor in the solution The numerator reacts to cause a change in the potential of the sensor. The change in potential is input as the original signal and is converted into a digital signal by A/D into the computer.
  • the computer performs raw data preprocessing and feature variable extraction successively.
  • the amount of information extracted by the three sensors is large, and in the acquisition of the original data, some noise signals are inevitably introduced due to external factors.
  • the original data is filtered and reduced in dimension by wavelet analysis, independent component analysis and principal component analysis, and the corresponding feature variables are extracted from the pre-processed information.
  • the color feature variables of the dry tea color and the texture feature variables of the dry tea shape are respectively extracted from the image information; the characteristic variables capable of characterizing the dry tea aroma are extracted from the olfactory sensor signal; and the tea sensor taste is extracted from the taste sensor signal. Characteristic variable.
  • these feature vectors are respectively correlated with the results of the expert sensory review in the database, and the sensor description of the quality indicators of the famous tea is realized by the machine learning method.
  • the computer simulates the human brain, synthesizes and balances these three kinds of information, and constructs a multi-information fusion model for intelligent evaluation of famous tea instruments through appropriate pattern recognition method, which is to describe the three kinds of sensor information separately.
  • the results of each quality index of tea were fused at the decision level, and then the nonlinear pattern recognition method was used to establish a multi-information fusion model for the intelligent review of famous tea instruments.
  • the sample to be tested is extracted by the corresponding data acquisition and feature vector, and the extracted feature vector is substituted into the pre-established model to intelligently review the quality of the unknown sample.
  • the invention has the beneficial effects that: the invention provides an intelligent evaluation method for a famous tea quality instrument based on multi-sensor information fusion, and combines three kinds of sensor information of computer vision, electronic nose and electronic tongue to achieve the intelligence of the instrument.
  • the judging in order to fully imitate the human sensory organs, including vision, smell and taste, judge all the quality indicators of famous tea, and integrate with the results of expert review to achieve the intelligent evaluation of the comprehensive quality of famous tea.
  • the invention incorporates bionics and multi-information Compared with the human sensory inspection, the evaluation results are consistent and highly automated.
  • the invention can improve the intelligent management level of the famous tea market, and has direct directness to regulate the order of the famous tea market and maintain the famous tea brand. Realistic meaning.
  • FIG. 1 is a schematic diagram of a technical scheme for intelligent evaluation of a famous tea quality instrument based on multi-sensor information fusion according to the present invention.
  • DETAILED DESCRIPTION OF THE INVENTION The present invention has universality for the intelligent evaluation of the quality of famous teas. However, due to the variety of famous teas, the present invention only provides an example for the intelligent evaluation of the quality of the famous specialty tea Biluochun of Jiangsu, and the evaluation of other teas.
  • the quality level of the tea sample can be tested by referring to the method of the embodiment, specifically for the quality index index of the tested tea sample, and establishing a new discriminant model. Example steps are shown in Figure 1.
  • Biluochun samples (the number of samples is greater than 100), and the quality indicators of the sensory review by professionals.
  • Three professional reviewers conducted a sensory review of each sample.
  • the sensory evaluation indicators mainly include four factors (dry tea color, shape, aroma and taste of tea soup, etc.).
  • the reviewer scores the factors one by one, the average of the scores of the three reviewers as the final score of the quality factor, and builds a standard database from the data obtained by the professional sensory review to establish the next intelligent evaluation model. service.
  • the enriched gas is drawn into the entire array of olfactory sensors by a micro pump, and the acquisition of the olfactory sensor signal data is started.
  • 2 Computer vision (visual sensor) data acquisition, the tea raw materials after the olfactory sensor data acquisition is evenly spread in a special glass container, and then placed in a closed light source box to start image data collection.
  • the taste sensor is immersed in the tea soup solution, at which time different functional groups on the sensor biofilm react with the odor molecules in the solution, causing a change in the potential of the sensor, and the change value of the potential is input as the original signal.
  • input to the computer by A/D conversion to digital signal.
  • the three sensors respectively collect sensor information reflecting the quality indicators of Biluochun, and input the computer through the corresponding data acquisition card.
  • the computer performs raw data preprocessing and feature variable extraction successively. The amount of information extracted by the three sensors is huge, and in the acquisition of the original data, some noise signals are inevitably introduced due to external factors.
  • the original data is subjected to noise filtering and dimensionality reduction preprocessing through wavelet analysis, independent component analysis and principal component analysis, and corresponding feature variables are extracted from the preprocessed information.
  • the color characteristic variables which characterize the color of Biluochun dry tea and the texture characteristic variables which characterize the shape of dry tea are extracted from the image information.
  • the characteristic variables which can be used to characterize the aroma of Biluochun dry tea are extracted from the olfactory sensor signal.
  • the characteristic variable of the taste of Biluochun tea soup are respectively correlated with the results of the expert sensory review in the database, and the sensor description of the quality indicators of Biluochun is realized by the machine learning method.
  • the computer is subjected to data preprocessing and feature variable extraction, and is associated with the results of the expert sensory review in the database.
  • the sensor's identity description of each quality indicator of Biluochun is realized.
  • the three kinds of information are integrated.
  • Balance construct the multi-information fusion model of Biluochun instrument intelligent review through pattern recognition method; it is to combine the three kinds of sensor information to describe the results of each quality index of famous tea at the decision level, and then establish a famous tea by nonlinear pattern recognition method.
  • the Biluochun sample to be tested is extracted by the corresponding data acquisition and feature vector, and the extracted feature vector is substituted into the pre-established model to intelligently evaluate the quality of the unknown Biluochun sample.

Landscapes

  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Food Science & Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Pathology (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Immunology (AREA)
  • Medicinal Chemistry (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Tea And Coffee (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Description

基于多传感信息融合的名优茶品质仪器智能化审评方法 技术领域 本发明涉及一种针对名优茶品质的仪器智能审评方法, 特指基于多传感信息融合的名 优茶品质仪器智能化审评方法。 背景技术 由于名优茶价格高,产品的利润空间大,一定程度上导致了名优茶市场存在以次充好、 以假乱真的现象。 名优茶产品鱼目混珠, 质量良莠不齐, 给消费者带来了购买风险, 不仅 损害了消费者的利益, 也不利于名优茶的品牌维护。 长期以来, 感官审评一直被视为检验 茶叶品质的最基本方法,通过训练有素的专业审评人员来审评的内容一般包含色、香、味、 形和叶底 (茶渣)等, 分别代表茶叶品质的不同方面。 但感官审评存在以下两个局限性: ①感官审评毕竟是一种专家行为, 而专家的培养是一个长期的训练和经验获取过程, 培养 一个训练有素的感官审评专家是要付出较大代价的, 普通消费者一般不具备这种能力; ② 审评受主客观因素干扰较大, 即使训练有素的高级审评专家, 他们感觉器官的灵敏度也受 到经验、 性别、 精神状态、 身体状况甚至地域环境等因素的千扰而改变, 从而影响感官审 评结果的准确性和稳定性。 因此, 研究一种快速便捷的名优茶品质智能评审方法对于规范 名优茶市场秩序, 维护名优茶品牌, 提高名优茶品质乃至振兴茶叶产业都有着直接的现实 意义。 理论研究表明, 计算机视觉技术、 电子鼻和电子舌技术可以用于茶叶品质检测和分类 鉴别。 经检索, 从检索结果看, 目前仅有一个计算机视觉鉴别茶叶类别方面的国内相关专 利 (实用新型), "基于多光谱图像技术的纹理分析鉴别不同绿茶的装置", 专利号:
ZL2007201100404.1 ;电子鼻和电子舌技术在茶叶品质检测上的应用还仅仅停留在实验室阶 段, 没有相关专利文献。 并且在茶叶品质检测上, 所采用的都是单一的检测手段, 具有一 定的局限性。 反映茶叶品质的指标是多方面的, 包括色、 香、 味、 形等感官指标, 而某种 单一的检测手段往往不能全面地描述一个对象, 只能描述其中的一个方面, 这种侧重点不 同带来的局限性必然影响到检测结果的精度和稳定性。 本发明提供一种基于多传感信息融 合的名优茶品质仪器智能化审评方法, 利用计算机视觉 (视觉传感器)、 电子鼻(嗅觉传 感器)和电子舌(味觉传感器)等三种传感器分别模拟人的眼、 鼻和舌三种感觉器官, 分 别采集能反映名优茶品质指标的各传感信息; 将这些特征向量分别与专家感官审评的结果 相关联, 通过适当的学习和训练, 实现传感器对名优茶品质指标的身份描述; 最后, 将这 三种信息进行综合与平衡, 通过合适的模式识别方法构建名优茶仪器智能审评的多信息融 合模型。 待测样本通过相应的数据采集和特征向量提取, 将提取的特征向量代入预先建立 好的模型就可以对未知样本的品质进行智能化审评。 本发明能模仿人的感官审评方法对名 优茶品质进行智能化审评, 审评结果一致性好、 自动化程度强, 为茶叶产品品质标准化分 级创造条件。 发明内容 鉴于上述现有技术发展情况, 为克服当前单一传感器检测技术在茶叶品质检测中存在 的局限性, 本发明的目的就是要提供一种基于多传感信息融合的名优茶品质仪器智能化审 评方法, 利用计算机视觉、 电子鼻和电子舌三种传感器分别模拟人的眼、 鼻和舌三种感觉 器官, 分别采集能反映名优茶品质指标的各传感信息; 将这些特征向量分别与专家感官审 评的结果相关联, 通过适当的学习和训练, 实现传感器对名优茶品质指标的身份描述; 最 后, 将这三种信息进行综合与平衡, 通过合适的模式识别方法构建名优茶仪器智能审评的 多信息融合模型。 待测样本通过相应的数据采集和特征向量提取, 将提取的特征向量代入 预选建立好的模型就可以对未知样本的品质进行智能化审评。 名优茶仪器智能评审模拟人 工感官评审的方法。 通过利用计算机视觉技术、 电子鼻和电子舌技术等多信息传感器分别 采集茶叶外形、 香味、 滋味和汤色信息数据, 再将这些信息数据相互融合与知识库中专家 知识经验相结合, 建立判别模型对茶叶的品质进行综合评判。 本发明基于多传感信息融合的名优茶品质仪器智能化审评方法按照下述步骤进行:
( 1 ) 针对某一种名优茶, 首先选取一批具有代表性的样本(样本数大于 100), 由专 业人员对名优茶各品质指标进行感官审评, 在审评结果的基础上建立一个标准数据库; 其 中所述标准数据库的建立, 由 3位以上专业审评人员对名优茶样本进行感官审评; 感官审 评指标主要包括 4个因子(干茶色泽、 形状、 香气和茶汤滋味等); 审评员对各因子逐一 感官评审打分, 专业审评人员得分的平均值作为该品质因子的最终得分, 通过专业人员感 官审评得到的数据构建一个标准数据库, 为下一步智能审评模型建立服务。
(2)利用计算机视觉(视觉传感器)、 电子鼻(嗅觉传感器)和电子舌(味觉传感器) 三种传感器分别采集能反映名优茶品质指标的各传感信息传入计算机: ①在嗅觉传感器数 据釆集过程中, 每次称取 10±0. 5g茶叶原料作为一个样本, 将其置于电子鼻系统的采样 杯中富集 10分钟, 通过微量泵将富集后的气体抽入电子鼻的传感器阵列, 开始电子鼻信 号数据的采集; ②在视觉传感器数据采集过程中, 将完成电子鼻数据采集后的茶叶原料均 匀地平铺在玻璃容器中, 然后将其置于自制光源箱内开始图像数据采集; ③在味觉传感器 数据采集过程中, 将完成图像数据采集的茶叶原料置于烧杯中, 加入去离子水, 在 60~80 °C条件下水浴 10~20min后过滤、 离心, 取上清液 5mL稀释 20倍, 移液到电子舌系统的烧 杯中进行电子舌数据采集;在进行电子舌数据采集过程中,将味觉传感器浸没茶汤溶液中, 此时传感器生物膜上面的不同官能团与溶液中的致味分子发生反应, 引起传感器电势的变 化, 电势的变化值作为原始信号输入, 通过 A/D转换成数字信号输入计算机。
(3 ) 计算机先后进行原始数据预处理和特征变量提取, 三个传感器提取的信息量庞 大, 并且在原始数据的采集中, 由于外界因素的影响, 不可避免地引入一些噪声信号。 先 通过小波分析、 独立分量分析和主成分分析等对原始数据进行滤噪和降维预处理、 再从预 处理后的信息中分别提取相应的特征变量。 其中从图像信息中分别提取表征干茶色泽的颜 色特征变量和表征干茶外形的纹理特征变量; 从嗅觉传感器信号中提取能表征干茶香气的 特征变量; 从味觉传感器信号中提取能表征茶汤滋味的特征变量。 然后, 将这些特征向量 分别与数据库中专家感官审评的结果相关联, 通过机器学习方法, 实现传感器对名优茶各 品质指标的身份描述。
(4)最后, 计算机模拟人的大脑, 将这三种信息进行综合与平衡, 通过合适的模式 识别方法构建名优茶仪器智能审评的多信息融合模型, 就是将三种传感信息分别描述名优 茶各品质指标的结果在决策级进行融合, 然后非线性的模式识别方法建立名优茶仪器智能 审评的多信息融合模型。 将待测样本通过相应的数据采集和特征向量提取, 将提取的特征 向量代入预先建立好的模型就可以对未知样本的品质进行智能化审评。 本发明的有益效果是: 本发明提供的是一种基于多传感信息融合的名优茶品质仪器智能化审评方法, 将计算 机视觉、 电子鼻和电子舌三种传感器信息融合起来达到仪器的智能评审, 来全面模仿人的 感官器官, 包括视觉、 嗅觉和味觉, 对名优茶各品质指标进行全方面的评判, 并与专家评 审的结果相融合, 以实现名优茶综合品质的智能化审评。 本发明融入了仿生学和多信息融 合思想, 与人的感观检测相比, 审评结果一致性好、 自动化程度强, 本发明可以提高名优 茶市场的智能化管理水平, 对于规范名优茶市场秩序, 维护名优茶品牌有着直接的现实意 义。 附图说明 图 1为本发明基于多传感信息融合的名优茶品质仪器智能化审评方法技术方案示意图。 具体实施方式 本发明对名优茶品质智能化审评具有通用性, 但由于名优茶种类很多, 因此本发明只 举一个用于江苏特产名优茶碧螺春品质智能审评的实施实例, 其它茶叶的审评可参照该实 施实例的方法, 具体针对所测茶叶样本的品质指标指标, 建立一个新的判别模型, 就可以 对茶叶样本的品质等级进行测试了。 实施实例步骤参阅图 1。 先选取一批不同质量等级碧螺春样本(样本数大于 100), 由 专业人员对其品质指标进行感官审评。 3位专业审评人员对每个样本进行感官审评。 感官 审评指标主要包括 4个因子 (干茶色泽、 形状、 香气和茶汤滋味等)。 审评员对各因子逐 一感官评审打分, 3个审评员得分的平均值作为该品质因子的最终得分, 通过专业人员感 官审评得到的数据构建一个标准数据库, 为下一步智能审评模型建立服务。 感官审评结束以后, 按照以下步骤进行数据采集: ①进行电子鼻(嗅觉传感器) 数据 采集, 每次称取 10±0. 5g茶叶原料作为一个样本, 将其置于电子鼻系统的采样杯中富集 10分钟,通过微量泵将富集后的气体抽入嗅觉传感器整列, 开始嗅觉传感器信号数据的采 集。 ②计算机视觉(视觉传感器)数据采集, 将完成嗅觉传感器数据采集后的茶叶原料均 匀地平铺在特制玻璃容器中,然后将其置于密闭光源箱内开始图像数据采集。③电子舌(味 觉传感器) 数据采集, 将完成图像数据采集的茶叶原料置于 200mL烧杯中, 加入 150mL 去离子水, 在 80'C条件下水浴 15min后过滤、 离心, 取上清液 5mL稀释 20倍, 移液到电 子舌系统的烧杯中进行电子舌数据采集。 在进行电子舌数据采集过程中, 将味觉传感器浸 没茶汤溶液中, 此时传感器生物膜上面的不同官能团与溶液中的致味分子发生反应, 引起 传感器电势的变化, 电势的变化值作为原始信号输入, 通过 A/D转换成数字信号输入计算 机。 三种传感器分别采集反映碧螺春各品质指标的传感信息, 并通过相应的数据采集卡输 入计算机。 计算机先后进行原始数据预处理和特征变量提取, 三个传感器提取的信息量庞 大, 并且在原始数据的采集中, 由于外界因素的影响, 不可避免地引入一些噪声信号。 先 通过小波分析、 独立分量分析和主成分分析等对原始数据进行滤噪和降维预处理、 再从预 处理后的信息中分别提取相应的特征变量。 其中从图像信息中分别提取表征碧螺春干茶色 泽的颜色特征变量和表征干茶外形的紋理特征变量; 从嗅觉传感器信号中提取能表征碧螺 春干茶香气的特征变量; 从味觉传感器信号中提取能表征碧螺春茶汤滋味的特征变量。 然 后, 将这些特征向量分别与数据库中专家感官审评的结果相关联, 通过机器学习方法, 实 现传感器对碧螺春各品质指标的身份描述。 计算机经过数据预处理和特征变量提取, 将其与数据库中专家感官审评的结果相关 联, 通过机器学习的方法, 实现传感器对碧螺春各品质指标的身份描述; 最后将这三种信 息进行综合与平衡, 通过模式识别方法构建碧螺春仪器智能审评的多信息融合模型; 就是 将三种传感信息分别描述名优茶各品质指标的结果在决策级进行融合, 然后非线性的模式 识别方法建立名优茶仪器智能审评的多信息融合模型。 将待测碧螺春样本通过相应的数据 采集和特征向量提取, 将提取的特征向量代入预先建立好的模型就可以对未知碧螺春样本 的品质进行智能化审评。

Claims

权利要求
1、基于多传感信息融合的名优茶品质仪器智能化审评方法,其特征在于按照下述步骤 进行: (1 )针对某一种名优茶, 首先选取一批具有代表性的样本, 由专业人员对名优茶各 品质指标进行感官审评, 在审评结果的基础上建立一个标准数据库; (2)利用计算机视觉 即视觉传感器、 电子鼻即嗅觉传感器和电子舌即味觉传感器三种传感器分别采集能反映名 优茶品质指标的各传感信息传入计算机; (3 )计算机先后进行原始数据预处理和特征变量 提取, 然后将这些特征向量分别与数据库中专家感官审评的结果相关联, 通过机器学习方 法, 实现传感器对名优茶各品质指标的身份描述; (4)最后, 计算机模拟人的大脑, 将这 三种信息进行综合与平衡, 通过模式识别方法构建名优茶仪器智能审评的多信息融合模 型; 将待测样本通过相应的数据采集和特征向量提取, 将提取的特征向量代入预先建立好 的模型就可以对未知样本的品质进行智能化审评。
2、根据权利要求 1所述的基于多传感信息融合的名优茶品质仪器智能化审评方法,其 特征在于其中所选取一批具有代表性的样本的样本数大于 100; 且由 3位以上专业审评人 员对名优茶样本进行感官审评; 感官审评指标主要包括干茶色泽、 形状、 香气和茶汤滋味 4个因子; 审评员对各因子逐一感官评审打分, 专业审评人员得分的平均值作为该品质因 子的最终得分。
3、 根据权利要求 1 所述的基于多传感信息融合的名优茶品质仪器智能化审评方法, 其特征在于在利用三种传感器分别采集能反映名优茶品质指标的各传感信息过程中: ①在 嗅觉传感器数据采集过程中, 每次称取 10±0. 5g茶叶原料作为一个样本, 将其置于嗅觉 传感器系统的采样杯中富集 10分钟, 然后将富集后的气体抽入嗅觉传感器阵列, 开始嗅 觉传感器数据采集; ②在视觉传感器数据采集过程中, 将完成嗅觉传感器数据采集后的茶 叶原料均匀地平铺在玻璃容器中, 然后将其置于自制光源箱内幵始图像数据采集; ③在味 觉传感器数据采集过程中, 将完成嗅觉和味觉传感器数据采集的茶叶原料置于烧杯中, 加 入去离子水, 在 60~80°C条件下 浴 10~20min后过滤、 离心, 取上清液 5mL稀释 20倍, 移液到电子舌系统的烧 中进行电子舌数据采集; 在进行电子舌数据采集过程中, 将味觉 传感器浸没茶汤溶液中, 此时传感器生物膜上面的不同官能团与溶液中的致味分子发生反 应, 引起传感器电势的变化, 电势的变化值作为原始信号输入, 通过 A/D转换成数字信号 输入计算机。
4、根据权利要求 1 所述的基于多传感信息融合的名优茶品质仪器智能化审评方法, 其特征在于在原始数据预处理和特征变量提取过程中, 先通过小波分析、 独立分量分析和 主成分分析对原始数据进行滤噪和降维预处理, 再从预处理后的信息中^^别提取相应的特 征变量; 其中从图像信息中分别提取表征干茶色泽的颜色特征变量和表征干茶外形的纹理 特征变量; 从嗅觉传感器信号中提取能表征干茶香气的特征变量; 从味觉传感器信号中提 取能表征茶汤滋味的特征变量。
5、 根据权利要求 1 所述的基于多传感信息融合的名优茶品质仪器智能化审评方法, 其特征在于构建名优茶仪器智能审评的多信息融合模型, 就是将三种传感信息分别描述名 优茶各品质指标的结果在决策级进行融合, 然后非线性的模式识别方法建立名优茶仪器智 能审评的多信息融合模型。
PCT/CN2009/001538 2009-10-09 2009-12-23 基于多传感信息融合的名优茶品质仪器智能化审评方法 WO2011041925A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN200910232916.9 2009-10-09
CN2009102329169A CN101692053B (zh) 2009-10-09 2009-10-09 基于多传感信息融合的名优茶品质仪器智能化审评方法

Publications (1)

Publication Number Publication Date
WO2011041925A1 true WO2011041925A1 (zh) 2011-04-14

Family

ID=42080762

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2009/001538 WO2011041925A1 (zh) 2009-10-09 2009-12-23 基于多传感信息融合的名优茶品质仪器智能化审评方法

Country Status (2)

Country Link
CN (1) CN101692053B (zh)
WO (1) WO2011041925A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104678843A (zh) * 2015-01-24 2015-06-03 福建农林大学 基于人体健康信息及空气成分数据的室内植物推荐方法
CN112006120A (zh) * 2020-08-27 2020-12-01 江南大学 基于气味及3d结构光的茶叶锅式杀青系统
CN113466409A (zh) * 2021-06-24 2021-10-01 上海应用技术大学 一种通过电子鼻来鉴定正山小种红茶的不同生产工艺的方法
CN114166770A (zh) * 2021-11-19 2022-03-11 安徽农业大学 一种基于纳米卟啉溶液相紫外可见光谱监测红茶发酵质量的方法
CN114814126A (zh) * 2021-10-30 2022-07-29 湖南农业大学 一种茉莉花茶的香气和滋味的品质量化的感官审评方法
CN117152648A (zh) * 2023-10-30 2023-12-01 泰州爱贝文化传媒有限公司 一种基于增强现实的辅助教学识图装置
CN117809124A (zh) * 2024-03-01 2024-04-02 青岛瑞思德生物科技有限公司 基于多特征融合的医学图像关联调用方法及系统

Families Citing this family (35)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102084794B (zh) * 2010-10-22 2013-03-06 华南农业大学 多传感器信息融合的作物病虫害早期检测方法及装置
CN102721793B (zh) * 2012-06-11 2015-04-22 江苏大学 一种食醋品质数字化检测方法及装置
CN103235087B (zh) * 2013-04-10 2015-10-28 江苏大学 一种基于嗅觉和味觉传感器信息融合的乌龙茶原产地的鉴别方法
CN103196954B (zh) * 2013-04-10 2015-08-26 江苏大学 一种基于嗅觉和味觉传感器信息融合的镇江香醋贮藏时间鉴别方法
CN103499609B (zh) * 2013-07-30 2016-01-20 中国标准化研究院 一种对蜂蜜香气智能嗅觉动态响应特征及差异化信息动态表征研究的方法
CN103472197B (zh) * 2013-09-10 2015-03-04 江苏大学 一种食品智能化仿生评价中的跨感知信息交互感应融合方法
CN104569328B (zh) * 2013-10-28 2017-10-20 云南天士力帝泊洱生物茶集团有限公司 普洱晒青茶滋味品质和香气品质的测定方法
CN105403507A (zh) * 2014-09-10 2016-03-16 中国农业科学院茶叶研究所 一种基于移动终端的龙井茶识别判断的系统及其方法
CN105866209B (zh) * 2016-03-28 2018-06-12 齐鲁工业大学 一种基于电子舌的大豆分离蛋白辨识方法
CN106227039B (zh) * 2016-08-24 2019-07-09 贵州铜仁和泰茶业有限公司 一种基于模式识别的茶叶加工设备控制方法
CN106172935A (zh) * 2016-08-26 2016-12-07 黄山王光熙松萝茶业股份公司 屯绿茶的拼配方法
CN106338517A (zh) * 2016-09-23 2017-01-18 江苏大学 基于视觉和嗅觉信息协同的水果新鲜度智能化评判方法
CN107886095A (zh) * 2016-09-29 2018-04-06 河南农业大学 一种基于机器视觉和嗅觉特征融合的分类识别方法
CN106353474B (zh) * 2016-10-10 2018-07-03 丽水学院 茶水鉴别装置及鉴别方法
CN106560702A (zh) * 2016-10-20 2017-04-12 中国计量大学 联合电子舌和色谱分离技术的武夷岩茶产地鉴别方法
CN106525849B (zh) * 2016-11-02 2019-03-05 江苏大学 茶叶智能化拼配方法与系统
CN108287161A (zh) * 2017-01-09 2018-07-17 中国计量大学 一种智能茶叶感官审评方法及系统
CN107220670B (zh) * 2017-05-27 2020-07-14 重庆大学 基于离散小波变换的有监督式人工味觉系统特征提取方法
CN107871016A (zh) * 2017-11-27 2018-04-03 胡明建 一种机械视觉和机械味觉以时间相互映射的设计方法
CN108981800B (zh) * 2018-06-25 2023-09-08 东北电力大学 一种利用神经动力学系统模型进行机器嗅-味联觉效应可视化的方法
CN108960315B (zh) * 2018-06-27 2021-08-06 河南农业大学 一种调理肉制品品质智能评价系统及方法
CN109189811B (zh) * 2018-08-09 2021-11-12 北京盈盛恒泰科技有限责任公司 水样识别方法及装置
CN109781949A (zh) * 2019-01-15 2019-05-21 江苏大学 一种基于视觉、嗅觉和味觉传感信息融合的茶叶原产地的判别方法
CN110018203A (zh) * 2019-03-18 2019-07-16 江苏大学 基于电子鼻电子舌智能感官技术的香醋风味定量表征方法
CN110133050B (zh) * 2019-04-10 2020-08-14 浙江大学 一种基于多传感器指纹图谱定性定量检测茶叶品质的方法
CN110133049B (zh) * 2019-04-10 2020-09-18 浙江大学 基于电子鼻和机器视觉的茶叶等级快速无损检测方法
CN109870548A (zh) * 2019-04-11 2019-06-11 广东省农业科学院茶叶研究所 一种英红九号红条茶品质的评鉴方法
CN110057975B (zh) * 2019-04-22 2023-07-25 广州德芯半导体科技有限公司 一种嗅感描述符获取方法、系统及相关装置
CN110307871B (zh) * 2019-07-11 2021-04-27 河南中医药大学第一附属医院 一种采用电子感官融合快速检测中药饮片质量的方法
CN111426801B (zh) * 2020-05-09 2022-08-02 上海宁和环境科技发展有限公司 电子鼻学习驯化方法及其设备
CN111693661A (zh) * 2020-06-11 2020-09-22 上海交通大学 基于智能感官的干腌火腿风味评价方法
CN111693658A (zh) * 2020-06-11 2020-09-22 上海交通大学 基于多种智能感官数据融合的食品品质鉴定方法
CN113970546A (zh) * 2021-09-03 2022-01-25 江苏大学 基于嗅觉味觉交互作用的绿茶品质可视化传感判别方法
CN114019100B (zh) * 2021-10-29 2024-03-26 中国农业科学院茶叶研究所 基于多源信息融合技术的滇红工夫茶汤综合品质客观量化评价方法
CN114398974A (zh) * 2022-01-11 2022-04-26 北京智进未来科技有限公司 一种基于多特征描述的茶叶品质审评方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE10315848A1 (de) * 2003-04-02 2004-10-14 Ahlers, Horst, Dr. Multisensor-Informationssystem
CN101059426A (zh) * 2007-05-29 2007-10-24 浙江大学 基于近红外光谱技术无损测量茶叶中茶多酚含量的方法
CN101419166A (zh) * 2008-11-18 2009-04-29 江苏大学 基于近红外光谱和机器视觉技术的茶叶品质无损检测方法及装置
CN101487825A (zh) * 2009-03-03 2009-07-22 中国农业大学 一种识别茶叶种类和/或等级的方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001327249A (ja) * 2000-05-19 2001-11-27 Kagoshima Prefecture 茶葉の評価方法及びその装置
CN1245630C (zh) * 2004-01-08 2006-03-15 江苏大学 基于电子视觉和嗅觉融合技术的农畜产品无损检测方法及其装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE10315848A1 (de) * 2003-04-02 2004-10-14 Ahlers, Horst, Dr. Multisensor-Informationssystem
CN101059426A (zh) * 2007-05-29 2007-10-24 浙江大学 基于近红外光谱技术无损测量茶叶中茶多酚含量的方法
CN101419166A (zh) * 2008-11-18 2009-04-29 江苏大学 基于近红外光谱和机器视觉技术的茶叶品质无损检测方法及装置
CN101487825A (zh) * 2009-03-03 2009-07-22 中国农业大学 一种识别茶叶种类和/或等级的方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHAO AIFENG ET AL.: "Application of Electronic Nose and Electronic Tongue in Tea Evaluation", FUJIAN AGRICULTURAL MACHINERY, no. 3, September 2007 (2007-09-01), pages 23 - 26 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104678843A (zh) * 2015-01-24 2015-06-03 福建农林大学 基于人体健康信息及空气成分数据的室内植物推荐方法
CN112006120A (zh) * 2020-08-27 2020-12-01 江南大学 基于气味及3d结构光的茶叶锅式杀青系统
CN112006120B (zh) * 2020-08-27 2022-05-31 江南大学 基于气味及3d结构光的茶叶锅式杀青系统
CN113466409A (zh) * 2021-06-24 2021-10-01 上海应用技术大学 一种通过电子鼻来鉴定正山小种红茶的不同生产工艺的方法
CN114814126A (zh) * 2021-10-30 2022-07-29 湖南农业大学 一种茉莉花茶的香气和滋味的品质量化的感官审评方法
CN114166770A (zh) * 2021-11-19 2022-03-11 安徽农业大学 一种基于纳米卟啉溶液相紫外可见光谱监测红茶发酵质量的方法
CN114166770B (zh) * 2021-11-19 2023-08-22 安徽农业大学 一种基于纳米卟啉溶液相紫外可见光谱监测红茶发酵质量的方法
CN117152648A (zh) * 2023-10-30 2023-12-01 泰州爱贝文化传媒有限公司 一种基于增强现实的辅助教学识图装置
CN117152648B (zh) * 2023-10-30 2023-12-26 泰州爱贝文化传媒有限公司 一种基于增强现实的辅助教学识图装置
CN117809124A (zh) * 2024-03-01 2024-04-02 青岛瑞思德生物科技有限公司 基于多特征融合的医学图像关联调用方法及系统
CN117809124B (zh) * 2024-03-01 2024-04-26 青岛瑞思德生物科技有限公司 基于多特征融合的医学图像关联调用方法及系统

Also Published As

Publication number Publication date
CN101692053B (zh) 2012-01-18
CN101692053A (zh) 2010-04-07

Similar Documents

Publication Publication Date Title
WO2011041925A1 (zh) 基于多传感信息融合的名优茶品质仪器智能化审评方法
CN103278609B (zh) 一种基于多源感知信息融合的肉品新鲜度检测方法
CN110133049B (zh) 基于电子鼻和机器视觉的茶叶等级快速无损检测方法
Nesselroade et al. Focus article: Idiographic filters for psychological constructs
CN110169770A (zh) 情绪脑电的细粒度可视化系统和方法
CN102023137B (zh) 一种白酒鉴别方法
CN102721793B (zh) 一种食醋品质数字化检测方法及装置
CN108960315A (zh) 一种调理肉制品品质智能评价系统及方法
CN102222164A (zh) 一种食品感官质量评价方法及系统
CN102735642B (zh) 一种快速无损鉴别初榨橄榄油和油橄榄果渣油的方法
CN102830087A (zh) 基于近红外光谱技术快速鉴别餐饮废弃油脂的方法
CN104182338B (zh) 一种疲劳驾驶预警产品检测精度的测试方法
CN104977258B (zh) 基于二维相关光谱的茶叶/化妆品等品质检测方法
CN110307871A (zh) 一种采用电子感官融合快速检测中药饮片质量的方法
CN108985584A (zh) 一种商标智能分析方法、装置、终端及存储介质
CN108287161A (zh) 一种智能茶叶感官审评方法及系统
CN106940292A (zh) 基于多光谱成像技术的虫蛀巴旦木原料快速无损鉴别方法
CN101685092A (zh) 用烟叶pH检测值判定烤烟醇化质量的方法
CN110414549B (zh) 一种模糊正交线性鉴别分析的茶叶近红外光谱分类方法
Shen et al. Discrimination of blended Chinese rice wine ages based on near-infrared spectroscopy
Dehan et al. Classification of Chinese Herbal medicines based on SVM
CN106570680A (zh) 用于评价浓香型原酒感官质量的方法
Lazaro et al. Chemometric data analysis for black tea fermentation using principal component analysis
CN202794093U (zh) 一种基于仿生嗅觉的烘焙食品质量快速检测装置
CN109614714A (zh) 一种发酵乳香气感官质量控制的模型建立方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 09850183

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 09850183

Country of ref document: EP

Kind code of ref document: A1