摘要
将能反映纹理空间尺度变化信息的尺度共生矩阵(动态信息)和反映纹理信息的灰度共生矩阵(静态信息)相结合,进行纹理特征抽取,对纹理图像进行分割,再对分割结果进行滤波,去除分割结果中存在的误分像素,结果表明,能够获得良好的分割效果.
This paper claims that the dynamic information-reflected scale co-occurrence matrix is combined with static information-reflected gray level co-occurrence matrix,and the texture images are segmented.In order to remove the misclassified pixels smoothing technique has been applied on the segmented images.The experiment shows that this method is effective in texture segmentation.
出处
《石家庄职业技术学院学报》
2007年第2期3-5,共3页
Journal of Shijiazhuang College of Applied Technology
关键词
尺度共生矩阵
灰度共生矩阵
纹理分割
滤波
KOHONEN神经网络
scale co-occurrence matrix
gray level co-occurrence matrix
texture segmentation
filter
Kohonen neural network
