基于 Sigmoid 核函数的纺织品色差分类检测研究
DOI:
作者:
作者单位:

作者简介:

通讯作者:

基金项目:

安徽省高校自然科学研究重点项目(KJ2019A0915)?


Research on Textile Color Difference Classification Detection Based on Sigmoid Kernel Function
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    颜色的自动检测分级是纺织、印染行业质量检测中的关键一环ꎮ 为达到纺织品颜色的快速分级ꎬ根据人类视觉特性ꎬ 提出了一种基于 Sigmoid 核函数的纺织品色差分类检测方法ꎮ 该方法首先将采集的待测纺织品图像进行预处理操作ꎬ并将图 像数据由 RGB 色彩空间转换至 HSV 色彩空间ꎻ其次对图像区域进行均匀分块ꎬ提取 H、S、V 分量值并采用加权和的方式计算 待测纺织品与标准样品间的色差值 ΔE ꎻ最后以 ΔE 作为特征向量ꎬ采用基于 Sigmoid 核函数的 SVM 分类器来实现纺织品颜色 分级ꎮ 通过色差检测系统分类实验验证该方法分类准确率较高ꎬ可以实现纺织品的色差检测分类ꎮ 研究结果可为陶瓷、木材 等其他行业的色差检测分类提供参考ꎮ

    Abstract:

    The automatic detection and classification of colors is a key part of the quality inspection in the textileꎬ printing and dyeing industries. In order to achieve rapid classification of textile colorsꎬ this paper proposes a textile color difference classification detection method based on the Sigmoid kernel function according to human visual characteristics. This method firstly preprocesses the collected textile images and converts the image data from RGB color space to HSV color spaceꎻ sec ̄ ondlyꎬ it categorizes the image areas and extracts H􀆯 S􀆯 and V component values and uses the weighted sum method to get the color difference value ΔE between the textile to be tested and the standard sample is calculated byꎻ finallyꎬ with ΔE as the eigenvectorꎬ the SVM classifier based on the Sigmoid kernel function is used to realize the textile color classification. The classification experiment of the color difference detection system verifies that the method has a high classification accu ̄ racyꎬ and can fulfill the color difference detection and classification of textiles. This article can also provide certain re ̄ search reference for the color difference detection classification of ceramicsꎬ timber and other industries.

    参考文献
    相似文献
    引证文献
引用本文

谈 波aꎬ郭家伟b ꎬ战惠惠aꎬ庞可染a.基于 Sigmoid 核函数的纺织品色差分类检测研究[J].西昌学院学报(自然科学版),2021,35(2):89-93.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2021-07-27