《基于数码相片的车载木材运输量自动识别系统设计与实现-毕业论文》由会员分享，可在线阅读，更多相关《基于数码相片的车载木材运输量自动识别系统设计与实现-毕业论文（36页珍藏版）》请在金锄头文库上搜索。
1、摘 要本文对数码相片车载木材自动识别进行研究，分析了车载木材运输量图像的预处理、特征提取等过程。车载木材自动识别可以对车载木材运输量进行自动识别，该技术运用于木材加工厂、交通运输管理等领域，能起到节省人力成本、提高效率、改进管理体系等作用。目前车载木材运输量识别系统好像已有产品投放市场，但这些产品大多依赖硬件设施的辅助才能达到较高的识别率。本论文主要是依靠软件来实现车载木材运输量的自动识别，使得识别系统具备更强的环境适应性。通过对收集的车载木材图像样本进行处理和分析，证明了本系统具有较好的实际应用性能。本论文对车载木材运输量自动识别系统的图像预处理，特征提取，边缘检测的实现三个部分进行了总结与
2、分析。论文首先总结了图像预处理技术和方法。图像预处理主要包括图像增强、图像分割、图像滤波等，图像增强主要是改善图像整体的视觉效果，图像分割是将图像中有意义或者需要的特征提取出来，滤波是在尽量保留图像细节特征的条件下对目标像的噪声进行抑制，其处理效果的好坏将直接响到后续图像处理和分析的有效性和可靠性。特征提取是图象处理的重要组成部分，论文主要介绍了主成分分析法、区域生长算法和边缘检测的方法，主成分分析法是用较少的变量去解释原来数据中的大部分变异，将许多相关性很高的变量转化成彼此相互独立或不相关的变量；区域生长算法的基本思想是将具有相似性质的像素点集合起来构成区域；边缘检测法在一定程度上大幅度地减
3、少了数据量，剔除了可以认为不相关的信息，保留了图像重要的结构属性。由于预处理后的木材图像整体颜色比较接近，主成分分析法和区域生长算法不适合作为其特征提取的方法，因此本文选取的特征提取的方法是边缘检测法。边缘检测的目的是在一幅图像中提取不同区域的轮廓，将图像分割成不同的区域，使每个区域都由大致相同的像素组成。前面预处理对木材图像进行滤波和锐化目的是为了提取更清楚的木材图像轮廓特征，在后面运用边缘检测法来提取木材图像轮廓特征，分别进行了Prewitt算子、Log算子Sobel算子的边缘检测，接着对木材图像进行了开闭运算，消除了边缘检测之后的边缘出现断点的现象，使得木材图像轮廓特征更加明显，便于自动
4、识别的识别。关键词：图像；木材识别；特征提取；边缘检测；模式识别-iii-ABSTRACTIn this paper, timber board automatic identification of digital photo research, analysis of timber traffic vehicle image preprocessing, feature extraction process. Automatic Vehicle Identification wood board timber traffic can be automatically identified,
5、 the technology used in wood processing plants, transportation management and other fields, can play a save on labor costs, increase efficiency, improve management system and function. The current volume of timber transportation vehicle identification system as existing products on the market, but t
6、hey are mostly dependent on the supporting hardware to achieve a higher recognition rate. This paper relies mainly on software to achieve the automatic vehicle identification timber traffic, making identification system has more to the environment. By collecting samples of vehicle timber image proce
7、ssing and analysis, proved the system has good practical application performance. The paper timber traffic on the board automatic identification system, image preprocessing, feature extraction, edge detection to achieve the three parts of a summary and analysis. Paper first summarizes the image prep
8、rocessing techniques and methods. Image preprocessing includes image enhancement, image segmentation, image filtering, image enhancement is mainly to improve the overall visual image, image segmentation is the image feature extraction of meaningful or need out of filtering is to preserve image detai
9、l features under the conditions of the target image noise suppression, the treatment effect is good or bad will have a direct bearing on the follow-up image processing and analysis of the validity and reliability. Feature extraction is an important part of image processing, the paper introduces the
10、principal component analysis, region growing algorithm and the edge detection method, principal component analysis method is less variable to explain most of the variation of the original data will be Many are highly relevant independent variables into each other or irrelevant variables; region grow
11、ing algorithm is the basic idea is to pixels with similar properties together constitute the regional; edge detection method to a certain extent, significantly reduce data volume in the deletion of relevant information can be deemed not to retain the important structural properties of the image. As
12、the pretreatment of wood close to the color image as a whole, the principal component analysis and region growing algorithm is not suitable for feature extraction method, this paper selected feature extraction method is the edge detection method. The purpose of edge detection is extracted in differe
13、nt regions of an image contour, image segmentation into different regions, so that each region formed by roughly the same pixel. Pretreatment of the wood in front of image filtering and sharpening the purpose of extracting more clearly outline the image characteristics of wood, behind the use of edg
14、e detection to extract the timber image contour features, respectively the Prewitt operator, Log Operator Sobel operator edge detection, then the image of the timber was open and close operation, eliminating the edge after edge detection breakpoint phenomenon occurred, making more obvious features o
15、f image contour timber, easy identification of automatic identification.Key Words：Image; wood identification; feature extraction; edge detection; pattern recognition湖南科技大学本科生毕业设计（论文）目 录第一章 绪论11.1 研究背景11.1.1 模式识别的含义11.1.2 模式识别的发展史21.1.3 模式识别方法21.1.4 模式识别的应用31.2 研究现状31.3 论文主要结构及内容4第二章 图像预处理62.1 图像增强62.2 图像分割62.2.1 全局阈值化72.2.2 自适应阈值72.3 图像滤波82.4 图像锐化82.5 本章小结8第三章 图像特征提取方法93.1 特征的含义93.2 常见的几种基本特征93.2.1 纹理特征93.2.2 形状特征113.2.3 空间关系特征113.3 图像特征提取的基本思路123.4 样本特征库初步分析133.5 图像特征提取的方法133.5.1 主成分分析法133.5.2 区域生长算法143.5.3 边缘检测法143.6 本章小结16第四章 木材图像特征提取研究174.1 线性高斯滤波174.1.1 高斯滤波的基本概念174.1.2 木材图像高斯滤波的实验结果和分析
