摘要
提出一种基于Canny算子的图像特征提取算法。首先对木材横截面放大图像进行统一缩放,利用Canny算子计算图像中主要的纹理特征和气孔的分布,然后计算图像的奇异值SVD,最后利用支持向量机SVM对奇异值进行训练,得到木材的分类器。实验表明Canny算子可以有效地取出图像中的噪点,提高SVD值得稳定性,木材识别的正确率得到极大地提高。
Presents a new image feature extracting method which based on canny algorithm. Firstly, takes pictures from the cross-section of wood and resize to union size, calculates the ray feature and distribution of pore. Secondly, calculates the singular value(SVD) of the image. Finally,classifies the wood species with SVD using the Support Vector Machine(SVM). The experiments show that the Canny algorithm can improve the accuracy of classification, it reduces the noise of the wood image and increases the robustness of classification.
引文
[1]McMillin Charles W. Application of Automatic Image Analysis to Wood Science[J]. Wood Science,1982,14(3):97-105.
[2]朱佳,汪杭军.基于Graph Cuts的木材扫描电镜图像特征提取方法[J].林业科学,2014,50(4):108-114.
[3]Khalid M,Yusof R,Khairuddin A S M. Tropical Wood Species Recognition System Based on Multi-Feature Extractors and Classifiers[C]. International Conference on Instrumentation Control&Automation,2012.
[4]Wang H,Qi H,Li W,et al. A GA-Based Automatic Pore Segmentation Algorithm[C]. ACM/Sigevo Summit on Genetic&Evolutionary Computation,2009.
[5]WANG Hang-jun,ZHANG Guang-qun,QI Heng-nian,LI Wen-zhu. A Review of Research on Wood Recognition Technology[J]. Journal of Zhejiang Forestry College,2009(6):896-902.
[6]LIU Zi-hao,QI Heng-nian,ZHANG Guang-qun,WANG Hang-jun. Wood Identification Method Based on Microstructure Images in Cross-Section. Scientia Silvae Sinicae,2013(11):1001-7448.
[7]张晶晶,范学良,蔡小舒.计算Canny算子双阈值的新方法.中国分体技术,2011(17):67-71.
[8]李晓东.基于子空间和流行学习的人脸识别算法研究[M].山东人民出版社,2013,6.
[9]Klema V,Laub A J. The Singular Value Decomposition:Its Computation and Some Applications[J]. Automatic Control IEEE Transactions on,1980,25(2):164-176.