麦冬、太子参、玉竹鉴定特征定量化及识别研究

发布时间：2018-05-29 05:29

本文选题：Matlab + 特征定量　；参考：《广州中医药大学》2017年硕士论文

【摘要】：目的:(1)结合计算机图像处理技术及摄像技术,定量测定不同产地的麦冬、玉竹、太子参的外观性状特征以及各种组织形态特征,同时构建所选药材不同产地的神经网络识别模型。利用图像处理软件,探究相应的算法,尽可能实现自动化、批量化操作,(2)对不同产地药材的总皂苷类成分进行组织化学定位和吸光度测定、建立组织特征参数与化学成分之间的相关性。为中药的鉴定及其质量标准控制研究提供一种新思路、新方法。方法:从所购得药材中随机抽取所需数量的样本,将所得样本置于白纸上,利用数码摄像设备摄取其宏观图片。而后,将已摄取的图片的药材按聚乙二醇包埋法制备显微装片,利用电子目镜摄像头在显微镜下无间隔地移动视野,且相邻视野间保持一定的重复区域,不间断摄取药材显微图像的全貌。利用Matlab平台编写程序结合Image Pro Plus图像测定软件对显微图像以及前面所摄取的外观图像进行处理,测量不同药材对应的宏观及显微特征参数。收集实验所需的各项参数并进行统计分析。对比同一药材不同产地或者是不同部位的特征差异,并由此构建出识别该类药材的特征向量。利用Matlab2014a自带的模式识别工具箱对药材的识别分析,同时采用相关分析对所选药材的特征与化学成分含量做统计分析结果:(1)实现了对麦冬、玉竹、太子参三种药材的形状、颜色、大小、表面纹理特征以及各组织特征的定量描述。(2)建立了适用于大量显微图像拼接的方法。(3)构建了识别麦冬、玉竹、太子参的神经网络模型。(4)建立了麦冬、玉竹、太子参三种药材的相应显微组织特征参数与化学成分含量的联系。(5)实现了麦冬药材的性状及组织特征的自动化提取。结论:本文基于显微摄像技术以及计算机图像处理技术的结合,较好地完成了麦冬、玉竹、太子参三种药材的常见鉴定特征的定量分析,同时构建了识别对应中药材的神经网络模型。该技术能更准确、客观地描述出麦冬、玉竹、太子参的传统鉴定特征,这为将来识别中药材的品种、产地等奠定基础。该技术结合化学成分与所测得鉴定特征的相关性分析,能够反映出中药的成分与其常见鉴定特征的联系,从而为从形态鉴定特征上评价和控制中药的质量提供了基础的依据。
[Abstract]:Objective: to determine quantitatively the appearance and morphological characteristics of Ophiopogon japonicus, Phyllostachys heterophylla and Radix Pseudostellariae in different habitats, combined with computer image processing technique and camera technique. At the same time, the neural network recognition model of the selected medicinal materials from different habitats was constructed. Using image processing software to explore the corresponding algorithm, realize automation and batch operation as far as possible, and carry out histochemical localization and absorbance determination of the total saponins of medicinal materials from different habitats. The correlation between tissue characteristic parameters and chemical constituents was established. It provides a new way and method for the identification and quality standard control of traditional Chinese medicine. Methods: the samples were collected randomly and placed on white paper. The macroscopical images were taken by digital camera. Then, the medicinal materials taken were prepared by the polyethylene glycol embedding method, and the electronic eyepiece camera was used to move the field of vision without interval under the microscope, and the adjacent field of vision kept a certain repeating area. Continuously take a full picture of the microscopic images of medicinal materials. The microscopic image and the appearance image taken in front were processed by using the program of Matlab and Image Pro Plus image measurement software, and the macroscopic and microscopic characteristic parameters of different medicinal materials were measured. Collect the parameters needed for the experiment and make statistical analysis. The characteristics of the same medicinal materials from different regions or different parts were compared, and the characteristic vectors were constructed to identify this kind of medicinal materials. The pattern recognition toolbox of Matlab2014a was used to identify and analyze the medicinal materials. At the same time, correlation analysis was used to analyze the characteristics and chemical composition of the selected medicinal materials. The results showed that the shapes and colors of Ophiopogon japonicus, Phyllanthus heterophylla and Radix Pseudostellariae were realized. Quantitative description of size, surface texture features and tissue characteristics. (2) A method for mosaic of a large number of microscopic images was established. A neural network model for identification of Ophiopogon japonicus, Phyllostachys heterophylla and Radix Ophiopogonis japonicus was established. The relationship between the microstructural characteristic parameters and the content of chemical components of Radix Pseudostellariae heterophylla was used to realize the automatic extraction of the characters and tissue characteristics of Radix Ophiopogonis. Conclusion: based on the combination of microscopic camera technique and computer image processing technology, the quantitative analysis of common identification characteristics of Radix Ophiopogonis, Phyllostachys heterophylla and Radix Pseudostellariae was completed in this paper. At the same time, a neural network model for identifying the corresponding Chinese medicinal materials was constructed. This technique can more accurately and objectively describe the traditional identification characteristics of Ophiopogon japonicus, Phyllostachys heterophylla and Radix Pseudostellariae, which will lay a foundation for the identification of varieties and habitats of Chinese medicinal materials in the future. This technique can reflect the relationship between the components of traditional Chinese medicine and their common identification characteristics, and provide the basis for evaluating and controlling the quality of traditional Chinese medicine from the aspect of morphological identification.
【学位授予单位】：广州中医药大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】