基于图像多小波变换的煤岩界面识别

基金项目:

北京市属高等学校创新团队建设提升计划资助项目（IDHT20130511）

中国矿业大学（北京）博士研究生拔尖创新人才培育基金资助项目（80015E639）

Coal-rock interface identification based on image multi-wavelet transformatio

1.
School of Information Engineering, China University of Geosciences(Beijing), Beijing 100083, China
2.
School of Electrical and Information Engineering, Beijing Polytechnic College, Beijing 100042, C

摘要: 针对现有煤岩界面识别方法存在普适性差、可靠性差等问题，提出了一种基于图像多小波变换的煤岩界面识别方法。对煤岩图像进行多小波变换，提取不同频带多小波系数在固定窗口大小下的标准差作为纹理测度并生成归一化多频带特征向量，利用朴素贝叶斯分类器对纹理特征进行识别。实验结果表明，对于分辨率为128×128的图像，当窗口大小为9，选择频带F5—F16构造特征向量时，识别率可达96.14%。
- 煤岩界面识别 /
- 多小波 /
- 多频带特征 /
- 朴素贝叶斯分类器
Abstract: In view of problems of poor universality and low reliability in existing identification method of coal-rock interface, an identification method of coal-rock interface based on image multi-wavelet transformation was proposed. Firstly, coal-rock image is transformed by multi-wavelet. Then standard deviation under fixed window size with multi-wavelet coefficients of different frequency bands is extracted as texture measure and normalization multi-band feature vector is formed. Finally, texture feature is identified by naive Bayes classifier. The experimental results show that identification rate can achieve 96.14% when window size is 9 and feature vector constructed by frequency bands 5-16 for image of resolution 128×128.
- coal-rock interface identification /
- multi-wavelet /
- multi-band feature /
- naive Bayes classifier