基于Gabor变换的木材表面缺陷图像分割方法

Image segmentation method of surface defects of wood based on Gabor transform

为了提高对木材表面缺陷图像分割的正确率,采用了环形Gabor滤波器将木材纹理图像变换到联合空间频率域,并在能量意义下定义了特征参数。根据多方向滤波结果形成缺陷图像的分割特征向量。结合模糊C均值聚类算法和数学形态学后处理操作提取出缺陷目标区域,分割正确率为98.29%。通过与基于灰度共生矩阵的分割方法进行比较实验,该方法平均分割精度比后者提高了4.22%,实验结果表明了该方法的可行性。

To improve the image segmentation accuracy of the wood surface defect, circular Gabor filter is used to convert texture image from airspace into the joint space frequency domain, and the characteristic parameters in energy sense are defined. According to the filtering results in multi-direction, segmentation feature vectors of defect images are collected. The target area of defects is drawn by combining the fuzzy C-means clustering algorithm and mathematical morphology post-processing, and the segmentation accuracy reaches to 98.29%. To be compared, the image segmentation experiments based on gray level co-occurrence matrix is conducted. The results show that the average segmentation accuracy of this method is higher by 4.22%, then the feasibility of the method is validated.