摘要
根据柳杉锯材表面的纹理特点,构造了适于描述柳杉锯材表面节疤缺陷和正常材纹理特征的空间灰度共生矩阵特征参数群,并优化选择了其中5个纹理特征参数用于区分柳杉锯材表面节疤缺陷和正常材.在统计了柳杉锯材表面死节、活节和正常材的5个纹理特征值分布范围的基础上,根据5个纹理特征值的分布范围及其对死节、活节和正常材的区分度,构建了柳杉锯材表面死节、活节和正常材的相应识别规则,基于图像纹理特征匹配技术开发了柳杉锯材表面节疤缺陷和正常材的自动识别系统.对于300个含有单个和/或复数个节疤缺陷(184个活节和156个死节)的柳杉锯材图像的自动识别试验结果显示,活节和死节的正确识别率分别为83.2%和90.4%,识别精度表明,基于图像纹理特征匹配技术对柳杉锯材表面的节疤缺陷进行自动识别是有效可行的.
An algorithm was proposed to detect the sound knots and dead knots on the sawn sugi lumber based on the grey level co-occurrence matrix (GLCM). Five parameters, which are reliable to indicate the differences of the textures among clear wood, sound knot and dead knot, were calculated from the GLCM. The recognizing rules were built according to the distributions of the five parameters. 300 wood defect samples were used to test the proposed system. The results showed that the correct recognition ratios for sound knots and dead knots were 83.2% and 90. 4% , respectively.
出处
《华南农业大学学报》
CAS
CSCD
北大核心
2009年第3期103-106,共4页
Journal of South China Agricultural University
基金
广东省自然科学基金(4400-E07116)
教育部留学回国人员科研启动基金
关键词
锯材缺陷
纹理特征
灰度共生矩阵
sawn lumber defects
texture features
gray level co-occurrence matrix