摘要
根据木材表面缺陷图像的自身特点,提出了基于灰度—梯度二维阈值向量的缺陷区域分割方法。该方法以灰度—梯度共生矩阵为模型,通过计算基于灰度—梯度共生矩阵的二维熵并使边缘区域的熵最大化来选择二维阈值向量。该方法不仅利用了图像的灰度信息,也利用了图像的梯度信息。采用形态学运算对分割后的二值图像进行分割后处理,试验表明,分割效果良好。
According to the characteristics of wood surface defects, an approach to the segmentation of defect region based on two-dimensional gray level-gradient threshold vector is presented. It takes gray level-gradient co-occurrence matrix as a model and uses maximum entropy theory, namely, the approach evaluates two-dimensional entropies based on gray level- gradient co-occurrence matrix, and 2D threshold vector which maximizes the entropies of edge region is selected. This method attempts to utilize the information both of gray level and gradient in an image. And it carries on post-processing to the segmented binary images by morphology operation. The segmentation experiment shows good results.
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2008年第9期53-55,共3页
Journal of Northeast Forestry University
基金
黑龙江省博士后基金项目(LBH-Z06032)
关键词
木材表面缺陷
图像分割
灰度一梯度共生矩阵
最大熵
形态学
Wood surface defects
Image segmentation
Gray level-gradient co-occurrence matrix
Maximum entro-py
Morphology
