摘要
针对基于骨架的单一特征的算法在异形纤维图像识别方面的不足,提出一种基于骨架特征并融合其他几何特征的层次化识别算法.首先采用轮廓跟踪算法将骨架信息映射到一种骨架树结构中,然后构造骨架特征邻接矩阵并求出该矩阵特征值,以特征值对异形纤维图像进行大类识别,再结合骨架周长统计特征和异形纤维图像轮廓的凹凸特征对每一大类异形纤维图像进行再分类,以达到精确识别的目的.研究结果表明算法对异形纤维图像具有良好的识别效果.
A hierarchical and skeleton-based algorithm which also combines with other geometrical features was proposed to improve the performance of profiled fiber image recognizer which use single skeleton feature.Firstly,the skeleton information was mapped to a skeleton tree structure by contour tracking algorithm.Then,a skeleton tree＇s adjacency matrix was constructed and eigenvalues of the matrix was worked out and the profiled fiber image was divided into several categories according to the eigenvalues.Finally,every category was re-identified by statistical feature of the skeleton perimeters and convex-concave feature of the fiber contour.The experimental results showed that the profiled fiber image can be accurately recognized by the proposed algorithm.
出处
《东华大学学报（自然科学版）》
CAS
CSCD
北大核心
2011年第3期356-361,共6页
Journal of Donghua University(Natural Science)
基金
中央高校基本科研业务费专项资金资助项目(B073)
关键词
异形纤维
骨架
骨架树
邻接矩阵
矩阵特征值
凹凸点
profiled fiber
skeleton
skeleton tree
adjacency matrix
eigenvalues
convex-concave point
