木材表面图像的缺陷分割与类型识别方法

为了识别死节、活节、虫眼三种木材表面缺陷,采用Gabor变换和模糊C均值聚类进行缺陷分割;采用数学形态学运算对分割图像进行了后处理;获取了木材缺陷区域的12维频率能量参数和2维几何形状参数;用支持向量机进行木材表面缺陷类型的识别。采用Gabor变换和模糊C均值聚类方法对死节、活节、虫眼三种木材表面缺陷的分割精度都达到94%以上,支持向量机对缺陷类型分类正确率达到93%以上,这说明本文的方法对木材表面缺陷的分割与识别是可行的。

The Method of Defects Segmentation and Recognition to Wood Surface Image

In order to recognize the wood surface defects of dead knot, live knot, and worm hole, Gabor transtorm and fuzzy C-means clustering algorithm were used to segment wood image defects. Mathematical morphology was used in post-processing operation of＂ segmented wood images, 12 frequency-enengy parameters and 2 shape parameters of defect targets were calculated. Support vector machines were used in the recognition of wood surface defect types.The segmentation accuracy to defects reached up to 94%, and the recognition accuracy to defect types of Support vector machines reached up to 93%. The result shows that it is feasible to segment and identify wood surface defects.