摘要蚕茧中称为“绵茧”的次下茧与“上茧”的品相特征差异相对较小，人工遴选需要技术经验极为丰富的人员承担，准确性易受评定者的主观因素和实际经验影响。
本文试图从蚕茧表面的纹理上提取两者的差异，文中依次介绍了蚕茧图像的获取、预处理、模式识别处理算法。图像获取利用超眼摄像机选定合适的焦距、拍摄区域进行拍摄。图像预处理包括图像灰度化和图像增强处理，图像增强主要是在频域中利用巴特沃斯高通滤波器进行锐化。图像识别模型采用了连通域算法和灰度共生矩阵模型，连通域算法是基于巴特沃斯高通滤波后的采用邻域的连通概念；灰度共生矩阵则是在灰度的统计模型下得到特征参数，并进一步采用了k近邻算法和基于遗传算法优化的BP神经网络进行数据处理；为了进一步提高正确率，将两种模型的三种算法的正确率进行了线性组合，建立了新的识别模式的组合模型。为了工程化应用，文章最后提出了一种基于表面皱缩纹理图像特征识别的蚕茧自动分拣系统的初步方案，给出了系统方案示意图。25730
关键词 上绵茧 图像识别 连通域 GLCM BP神经网络
毕业论文设计说明书外文摘要
Title Research on the technology of silkworm recognition based on the texture of shrink wrinkle
Abstract
The characteristic differences of “Floss”and “cocoon” is relatively small, artificial selection requires highly experienced technical staff commitment, susceptible to assess the subjective factors and practical experience.
This paper attempts to extract from the difference between the two surface texture cocoon, the paper successively introduced cocoon image acquisition,preprocessing, pattern recognition processing algorithms. The image capture is made by using the supereyes camera to select the appropriate focal length and the shooting area. Image preprocessing including image gray and image enhancement, which in the frequency domain is mainly the use of Butterworth high-pass filter sharpening. Image recognition model uses a connected domain algorithm and GLCM model, the former is based on the concept of using neighborhood connectivity Butterworth high-pass filtered; GLCM is characteristic parameters obtained in the statistical model gray and further use of the k nearest neighbor algorithm and optimization of BP neural network based on GA for data processing; in order to further improve the accuracy,which of the three algorithms were linear combination, a new recognition pattern Combined Model. For engineering applications, the article concludes with a preliminary scheme based on surface wrinkled texture image feature recognition cocoon automatic sorting system, the program gives a schematic diagram of the system.
Keywords cocoon ,Image recognition ,Connected domain, GLCM ,BP neural network
目 次
1 前言 1
1.1 选题背景及意义 1
1.2 国内外研究现状 1
1.3 研究内容和研究方法 2
2 蚕茧表面图像获取 4
2.1 图像获取装置 4
2.2 拍摄条件选择 5
3 蚕茧图像预处理 8
3.1 灰度化处理 8
3.2 图像增强 9
4 蚕茧表面纹理图像特征识别 12
4.1 连通域算法 12
4.2 灰度共生矩阵模型 14
4.3 k近邻算法 16
4.4 基于遗传算法的BP神经网络 18
4.5 组合模型 22 基于缩皱纹理的绵蚕识别技术研究+MATLAB程序:http://www.youerw.com/jixie/lunwen_19647.html