摘要
异型纤维自动检验过程中,传统边沿检测方法,受到图片质量的影响较大。而尺度不变特征变换(SIFT),直接从原始的灰度图像中提取特征信息,相对受图片质量的影响较小。本文提出了一种基于尺度不变特征变换以及支持向量机算法的纤维识别方法,实现了一个以该理论为基础的纤维自动识别系统,并获得了较为理想的识别结果。验证了尺度不变特征变换算子具有较强的稳定性、抗噪性、仿射不变性等特性。
In the process of shaped fibers inspection, the traditional method of the edge detection is influenced by the quality of the images. Scale Invariant Feature Transform （SIFT） based on the original images set is less influenced by the quality of the images. And the characteristics of SIFT are invariable,which are the translation, rotation and affine of image in the scale space. In the paper,a fiber recognition method based on SIFT and SVM （Support Vector Machine） is proposed. Sequentially, an automatic fiber recognition system is built. The experimental result shows that the method is effective.
出处
《电子测量技术》
2009年第2期135-139,共5页
Electronic Measurement Technology
基金
全国优秀博士学位论文作者专项资金资助项目(200350)
教育部留学回国人员科研启动基金资助项目
关键词
尺度不变特征变换
局部不变特征
尺度空间
支持向量机
scale invariant feature transform
local invariant descriptor
scale space
support vector machine
