基于图像处理技术的棉纤维长度测量及其分布研究

发布时间：2018-05-31 15:12

本文选题：棉纤维 + 长度分布　；参考：《东华大学》2017年硕士论文

【摘要】：棉纤维长度指标是棉花质量重要的评价指标之一,对纺纱工艺设计和纱线质量有显著影响。棉花是重要的经济农作物,棉纤维检测技术提高有利于改善棉花品种和调整种植结构,增加棉农收入。由于目前测量方法存在测试速度慢、测试原理不明、测试结果不准等弊端。因此,本文在总结当前棉纤维长度测量的现状,分析当前研究工作的不足的基础上,提出基于图像处理技术的棉纤维长度测量方法,进一步深入研究棉纤维长度及分布的测试与分析。论文制定了课题研究的总体设计方案,主要包括棉纤维的制样、图像的采集和处理及实验验证、棉纤维长度分布与长度指标的测量、与其他测试系统结果的对比分析、最佳取样量的研究分析。论文完成的主要工作如下:(1)采用微量点取样法进行取样,人工将棉纤维逐根排列在黑色卡纸上进行纤维分离,使用扫描仪采集棉纤维图像。对图像进行图像转换、增强处理、二值化处理以及形态学处理,得到符合特征提取的图像。通过逐根对棉纤维图像的像素进行统计和换算,得到棉纤维长度。采用30mm切段涤纶长丝和新疆生产的细绒棉验证了本课题基于图像处理技术的棉纤维长度测量方法的可行性。(2)采用论文方法测得棉纤维长度原始数据进行整理得到的平均长度、上半部平均长度要比Y111、KX730、HVI、AFIS的结果长1~2mm;棉纤维长度整齐性结果与Y111较为接近,比HVI、KX730要差,比AFIS要好;棉纤维短绒率测试结果与Y111测试结果十分接近,分别为5.91%和5.80%,与其他测试方法相差较大,尤其是与AFIS的根数短绒率相差高达7.44%。这里最主要的原因应该是本课题测量的棉纤维未经梳理,几乎没有损伤。(3)针对两种利用图像处理技术测量的棉纤维长度数据,分别随机抽取N根纤维长度数据。利用相关系数、残差平方和、面积吻合度三个指标,评价N根纤维长度分布曲线与总体分布曲线吻合程度。通过分析两种棉纤维的指标数据折线图,得到本课题测量方法的最佳取样量范围为3000根~3500根。(4)采用MATLAB软件进行基于图像处理技术的棉纤维长度测量方法软件开发与设计,软件可方便快捷地测量棉纤维长度和清晰直观地呈现测量结果。利用细度为1.71dtex的细绒棉验证了软件的可行性,求得该细绒棉的长度分布与长度指标,和长度分布曲线吻合度评价指标结果。本文基于图像处理技术的棉纤维长度测量方法能够准确测量棉纤维长度,得到棉纤维长度分布和长度指标。在棉纤维长度测量研究中本课题研究结果具有一定的参考价值,为今后棉纤维长度快速测量奠定一定的基础。
[Abstract]:Cotton fiber length index is one of the important evaluation indexes of cotton quality, which has significant influence on spinning process design and yarn quality. Cotton is an important economic crop. The improvement of cotton fiber detection technology is helpful to improve cotton varieties, adjust the planting structure and increase the income of cotton farmers. The current measurement method has some disadvantages, such as slow test speed, unclear testing principle and inaccurate test results. Therefore, on the basis of summarizing the current situation of cotton fiber length measurement and analyzing the deficiency of current research work, this paper puts forward a method of cotton fiber length measurement based on image processing technology. The measurement and analysis of cotton fiber length and distribution were further studied. The overall design scheme of the research is made, including the sample preparation of cotton fiber, the collection and processing of image and the experimental verification, the measurement of the length distribution and length index of cotton fiber, and the comparison and analysis with the results of other test systems. Research and analysis of optimum sampling amount. The main work of this paper is as follows: (1) sampling with micro-point sampling method, the cotton fiber is arranged one by one on black paper to separate the fiber, and the scanner is used to collect the image of cotton fiber. The image is transformed, enhanced, binary and morphologically processed, and the feature extraction image is obtained. The length of cotton fiber is obtained by counting and converting the pixels of cotton fiber image one by one. The feasibility of the method of cotton fiber length measurement based on image processing technology was verified by using 30mm cut polyester filament and fine velvet cotton produced in Xinjiang.) the average length of cotton fiber length was obtained by using the method of paper to measure the raw data of cotton fiber length. The average length of the upper half is 1mm longer than that of Y111KX730 HVIAFIS, and the neatness of cotton fiber length is close to Y111, worse than that of HVIKX730 and better than that of AFIS, and the test results of cotton staple ratio are very close to those of Y111. It was 5.91% and 5.80% respectively, which was different from other test methods, especially with AFIS. The main reason should be that the cotton fiber measured in this paper is not combed and has almost no damage. Aiming at two kinds of cotton fiber length data measured by image processing technology, N fiber length data are randomly selected. The correlation coefficient, the sum of squared residuals and the area coincidence degree were used to evaluate the consistency between the N fiber length distribution curve and the total fiber distribution curve. By analyzing the curve diagram of the index data of two kinds of cotton fibers, the best sampling range of the method is 3000. The software of cotton fiber length measurement based on image processing technology is developed and designed by MATLAB software. The software can measure the length of cotton fiber and show the result clearly and intuitively. The feasibility of the software is verified by using the fine velvet cotton with fineness of 1.71dtex. The length distribution and the length index of the fine velvet cotton are obtained, and the evaluation results of the consistency of the length distribution curve are obtained. In this paper, the cotton fiber length measurement method based on image processing technology can accurately measure the cotton fiber length, and obtain the cotton fiber length distribution and length index. The results of this paper have certain reference value in the research of cotton fiber length measurement, and lay a certain foundation for the fast measurement of cotton fiber length in the future.
【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】：TS117;TP391.41

【参考文献】