节点文献

基于图像的皮革颜色纹理检测分类方法研究及系统研制

Research on Classifying Methods and Development of System Based on Leather Color and Texture

【作者】 宋春刚

【导师】 王文; 陈子辰;

【作者基本信息】 浙江大学 , 机械制造及其自动化, 2006, 硕士

【摘要】 近年来,随着计算机视觉技术的发展和在各个领域的广泛应用,使得皮革行业在这方面也得到了迅猛的发展。为适应社会和市场的发展需求,提高皮革产品的科技含量及生产效率,实现自动化与智能化生产,利用计算机对皮革制品图像进行自动分类识别的研究和应用显得尤为重要。本文主要研究皮革制品颜色和纹理分类识别方法,并在此基础上,研制开发计算机检测和分类系统,以达到应用要求。 本论文主要研究内容有: 一、分析皮革颜色和纹理特性。主要从皮革表面颜色和纹理一般表现特征进行统计研究,归纳出一般皮革制品的颜色和纹理特征情况,为本文后续的皮革颜色和纹理分类方法研究作下铺垫。 二、开展皮革颜色分类识别方法的研究,提出运用主色聚类和直方图相结合的分类识别方法可以快速、稳定的进行分类,同时对颜色空间模型和一般直方图提出改进,采用累加直方图消除颜色量化误差,最后给出了实验验证。 三、根据皮革纹理分析特性,运用相对极值的分类方法,并结合图像网格法,对皮革制品进行水平和垂直两个方向的灰度值扫描,提取出纹理特征值,并由此定义纹理粗糙度和深刻度两个参量来定义皮革纹理类别,从而达到纹理分类的目的。 四、以颜色和纹理识别技术为基础,开展了皮革图像检测和分类系统的研制,主要包括对硬件设备的开发和对软件系统的功能设计,并对每个模块设计思想和实现原理进行详细阐述。 五、通过应用实例,对皮革检测分类系统的具体应用流程以及实际应用效果进行了详细阐述,验证了系统可行性。

【Abstract】 With the spread application of computerized vision technology in various fields, these years have witnessed great development of leather industry in this aspect. To improve productivity of leather product, and to achieve automatic and intelligent production, it is important to classify leather products images with computer. The paper focused mainly on classification methods of leather colors and texture. Moreover, the author developed computerized inspection and classification system for application. The contents of the paper include:1. Analysis of leather product color and texture. This research generalized characteristic of color and texture on leather goods, based on the features of leather surface, which provided basis for the next study.2. To classify leather colors and texture quickly and steadily, the paper put forward the classification method combing main color clustering with color histogram. Meanwhile, the paper made some improvement in the methods of color histogram and used accumulative histogram to eliminate the digitized errors of color, which was verified by experiment.3. According to characters of leather texture, the paper applied a method of scanning both vertically and horizontally to derive eigenvalue of texture to define coarseness and profundity for measuring leather texture.4. Based on the identification technology of leather color and texture, the paper developed images inspection and classification system. It mainly included the design of hardware equipment and software system. In the rest of the section, elaborated design and principle of each module were given.5. The paper introduced practical application of leather inspection and classification system, and the effect was represented as well.

  • 【网络出版投稿人】 浙江大学
  • 【网络出版年期】2007年 01期
  • 【分类号】TS51
  • 【被引频次】14
  • 【下载频次】562
节点文献中: