基于灰色系统理论的显微纤维边缘检测系统研制

发布时间：2017-12-26 22:34

本文关键词：基于灰色系统理论的显微纤维边缘检测系统研制　出处：《东华大学》2017年硕士论文　论文类型：学位论文

【摘要】：伴随经济技术的飞跃,纤维在社会各领域作用凸显。纤维的成分及质量决定了其性能和价格,而检测出的纤维边缘的连续性与完整性又对纤维特征提取、目标分离、模式识别及计算混纺比的准确度影响很大。故纤维图像的边缘检测是纤维自动识别体系中的关键一环。然而纤维在包埋、切片以及采样过程中受到不均匀光照、聚焦、噪声等方面的干扰,无法避免的出现显微纤维图像质量差、背景区域与目标纤维区分不明显、轮廓不完整或者粘连等现象。而且传统边缘检测算法获取显微纤维边缘出现虚假、不连续边缘和不能有效提取粘连纤维边缘的问题。此外,出入境检验检疫局作为我国进出口检验验定的重要机关,一直存在纤维成分检测的相关问题,即虽耗费一定的人力物力,但却无法实现相对准确的检测及鉴证。故研究基于计算机数字图像处理技术,结合科学的灰色系统理论开展显微纤维边缘检测是一项十分有意义的课题。本课题来源于国家自然科学基金面目项目《基于约束条件的非负矩阵分解算法及其在纤维自动识别中的应用研究》(61472075)。作为在纺织、纤维和面料等学科上在国内乃至世界上有极高学术地位的特色鲜明的高校,我校在棉麻以及化学纤维混纺领域研究成果积累颇丰。本文将探究初涉图像边缘检测领域的灰色系统理论与现行的数字图像处理技术相结合,把提取出连续、完整和准确的显微纤维边缘作为研究的核心。本文综合运用灰色系统理论、NIBLACK算法和孔洞填充算法,提出了一套基于灰色系统理论的显微纤维边缘检测技术路线,研制了显微纤维边缘自动检测平台系统。以显微纤维图像中各像素点为研究对象,基于灰色系统理论的边缘检测方案通过把GM(1,1)模型与NIBLACK算法相结合取得显微纤维的强弱边缘信息并实现有效的强弱边缘连接再运用孔洞填充算法来处理虚假边缘以提取出完整的边缘信息。针对基于灰色系统理论的显微纤维边缘检测系统设计,从算法原理入手,充分利用灰色系统理论在我国学术研究体系中的强势地位和先进水平,发挥了灰色系统理论的“少数据贫信息”的建模特点,展示了设计结果。实验结果表明,该算法能够提取出精确连续完整的显微纤维边缘。
[Abstract]:With the leap of economy and technology, fiber plays an important role in all fields of society. The composition and quality of fiber determine its performance and price, and the continuity and integrity of fiber edge has great influence on the accuracy of fiber feature extraction, target separation, pattern recognition and computational blending ratio. Therefore, the edge detection of fiber image is a key link in the automatic fiber recognition system. However, fiber is disturbed by uneven illumination, focusing, noise and so on during embedding, slicing and sampling. It can not avoid the phenomenon of poor quality of micro fiber image, indistinct distinction between background area and target fiber, incomplete contour or adhesion. Moreover, the traditional edge detection algorithm can obtain the false and discontinuous edges of the micro fiber edges and can not effectively extract the edges of the adhesive fibers. In addition, the entry exit inspection and Quarantine Bureau, as an important organ of import and export inspection and verification, has been related to the detection of fiber components, that is, though it costs a lot of manpower and material resources, it can not achieve relatively accurate detection and verification. Therefore, it is a very meaningful topic to study the edge detection of micro fiber based on the computer digital image processing technology and the scientific grey system theory. This project is based on the National Natural Science Foundation of China "constraint based non negative matrix factorization algorithm and its application in automatic fiber recognition" (61472075). As in the textile, fiber and fabric on the subject at home and abroad have very high academic status and distinctive features of the University, our school has accumulated in cotton and chemical fiber research. This paper will explore the new field of image edge detection based on grey system theory and digital image processing technology of the combination of the extracted continuous, complete and accurate micro fiber edge as the core. Based on the grey system theory, NIBLACK algorithm and hole filling algorithm, a set of gray fiber system edge detection technology roadmap is proposed. The automatic detection platform system of micro fiber edge is developed. For each pixel in the image of micro fiber as the research object, the theory of grey system through the edge detection scheme based on GM (1,1) model and NIBLACK algorithm combining edge information obtained micro fiber and realize effective edge connection and then use the hole filling algorithm to handle false edges to extract the complete edge information. According to the design of micro fiber edge detection system based on grey system theory, starting from the principle of the algorithm makes full use of grey system theory in the system of academic research in China in a strong position and advanced level, play a grey system theory "less data poor information" model, showing the results of the design. The experimental results show that the algorithm can extract accurate and continuous complete microfiber edges.
【学位授予单位】：东华大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：N941.5;TP391.41

【参考文献】