基于SEM图像的低维纳米材料自动分类方法

Automatic Classification Method for Low-Dimensional Nanomaterials Based on SEM Image

针对传统方法在低维纳米材料形貌检测和分类鉴别方面的不足,提出了一种基于扫描电子显微镜(SEM)图像的低维纳米材料自动分类方法.以纳米材料的SEM图像为基础,利用小波包分解技术对材料表面纹理特征进行提取,通过将纹理特征与支持向量机(SVM)相结合,实现了纳米材料的自动分类.该方法具有检测速度快、精度高、无损耗等诸多优点,可用于纳米材料大规模生产中的自动检测.对16种不同类别材料的SEM图像仿真结果表明,该方法的分类精度能够达到93.75%,证明了其在实际工程中的有效性.

In order to overcome the deficiencies of traditional morphology detection methods in classifica- tion and identification of low-dimensional nanomaterials, a novel low-dimensional nanomaterial automatic classification method based on scanning electron microscope （SEM） image was proposed. Based on the SEM images of nanomaterials, the texture features of the material surfaces were extracted by wavelet pack- et decomposition. Besides, the nanomaterials were classified by combining the texture features with sup- port vector machine（SVM）. Due to many advantages like fast detection speed, high precision and no loss,the method can be used for automatic detection of nanomaterials in large-scale production. The ex- periments with 16 kinds of nanomaterials indicate that the classification accuracy rate is 93.75% , thus the validity of this method in practical engineering is verified.