基于神经网络的中药材显微图像处理与分析

It is particularly important to distinguish,identify and control the chinese herbal medicine,because the degradation,fake shoddy and other factors of chinese herbal medicine affect its quality and clinical curative effect.The traditional identification mainly rely on subjective experience, the error rate of recognition is high.Physical and chemical identification must utilize special analysis instruments, which can make the management and operation complicated. the identification and testing of chinese herbal medicine based on the information technology is the new direction of development. From the pulse coupled neural network is suitable for processing biological tissue class image,this project intends to analyse the stereology characteristic of the microscopic image of the chinese herbal medicine, explore the pulse coupled neural dynamics rule,and then establish a more complete model of the pulse coupled neural network for processing the chinese herbal medicine microscopic image on the basis of preliminary studies.We will effectively combine the model of the pulse coupled neural network with other novel nonlinear processing methods, to build the image database of the chinese herbal medicine, to form the intelligent expert system of the processing and analysis of the chinese herbal medicines microscopic image with the functions of the image target segmentation, feature extraction, recognition and retrieval by using the model of the pulse coupled neural network.This system will provide a basis for the quality testing efficiency of the chinese herbal medicine, and a new way of the modernization detection and analysis to chinese herbal medicine.

由于中药材品质的退化及假冒劣质品的市场混入等因素影响了中药材的质量和临床疗效，对中药材识别、鉴定与控制显得尤为重要。传统中药材识别鉴定主要依赖主观经验，识别错误率高；现代理化鉴定要借助专门分析仪器，管理及操作复杂；因此基于现代信息理论技术的中药材处理、识别、检测及鉴定是新的发展方向。 本申请项目拟在前期研究的基础上，从脉冲耦合神经网络适合处理生物组织类图像的角度出发，分析中药材显微图像体视学特征，探索脉冲耦合神经元动力学规律，构建更为完善的脉冲耦合神经网络中药材显微图像处理模型；有效结合其它新颖的非线性处理方法，建立中药材图像信息库，搭建实现脉冲耦合神经网络图像目标分割、特征提取及识别检索等功能的中药材显微图像处理与分析智能专家系统，为提高中药材质量的检测效率奠定基础，同时，为中药材检测与分析的现代化提供一种新的途径。

传统中药材在新时代下的数字信息化处理是传承与发展祖国中医药文化面临的重要任务之一。本项目以获取药典收载分布在甘肃及其他省份的部分中药材显微图像作为研究对象，从中药材显微图像体视形态学特征出发，探索了脉冲耦合神经元处理图像信息的原理，构建了适合处理显微组织类的二维及三维脉冲耦合神经网络（PCNN）模型；研究了基于模糊集的中药材显微图像增强；针对图像不同噪声的污染，分别研究了基于2D-PCNN中药材显微图像脉冲噪声滤除、基于全变分（TV）的中药材显微图像复原；为有效提取中药材彩色显微图像的目标信息，提出了基于改进3D-PCNN的图像自动分割方法；针对显微图像特征提取与识别检索，分别研究了基于改进2D-PCNN和改进3D-PCNN的中药材图像特征提取与识别检索。通过对一系列处理算法的分析研究，初步实现了信息处理理论技术与生物医药的有效结合，提高了中药材信息化处理的客观性、准确性、可重复性和智能化程度，为中药材现代化分析与处理提供了一种新的途径。