欢迎访问林业科学,今天是

林业科学 ›› 2013, Vol. 49 ›› Issue (11): 116-121.doi: 10.11707/j.1001-7488.20131116

• 论文与研究报告 • 上一篇    下一篇

基于横切面微观构造图像的木材识别方法

刘子豪1, 祁亨年1, 张广群1, 汪杭军2   

  1. 1. 浙江农林大学信息工程学院 临安 311300;
    2. 浙江农林大学天目学院 临安 311300
  • 收稿日期:2012-12-19 修回日期:2013-06-21 出版日期:2013-11-25 发布日期:2013-11-26
  • 通讯作者: 汪杭军
  • 基金资助:

    国家自然科学基金项目(60970082);浙江农林大学人才启动项目(2013FR059);浙江农林大学研究生科研创新项目(3122013240224)。

Wood Identification Method Based on Microstructure Images in Cross-Section

Liu Zihao1, Qi Hengnian1, Zhang Guangqun1, Wang Hangjun2   

  1. 1. School of Information Engineering, Zhejiang A & F University Lin'an 311300;
    2. Tianmu College, Zhejiang A & F University Lin'an 311300
  • Received:2012-12-19 Revised:2013-06-21 Online:2013-11-25 Published:2013-11-26

摘要:

提出一种基于核主成分分析(KPCA)和自适应增强(AdaBoost)的木材识别算法。通过把图像投影到KPCA高维空间,利用PCA方法对该空间中的数据进行特征提取和压缩,使用Gentle AdaBoost进行分类。结果表明:本方法对基于横切面微观构造图像的木材识别,具有较高的识别率和算法鲁棒性且运行时间快的特点。

关键词: 核主成分分析, 自适应增强, 图像压缩, 木材识别, 计算机视觉

Abstract:

In this paper, a new method based on kernel principle component analysis(KPCA) and AdaBoost was proposed for wood identification. After wood images projecting into a high-dimensional space of KPCA, PCA method was used to extract features and compress those features. Then these well-prepared features were classified with Gentle AdaBoost. The experimental results showed that our method based on microstructure images in cross section had some good performances, such as higher discrimination, robustness and efficiency in running time.

Key words: kernel principle component analysis(KPCA), AdaBoost, image compression, wood identification, computer vision

中图分类号: