岩土材料微结构图像预处理方法  

Preprocessing methods for microstructural images of geomaterials

作　　者：胡昕[1,2] 洪宝宁[1] 孙秋 王伟[4] 

机构地区：[1]河海大学岩土工程科学研究所,江苏南京210098 [2]中国电力工程顾问集团西北电力设计院,西安710065 [3]南京市公路管理处公路科学研究所,江苏南京210012 [4]南京工程学院建筑工程学院,江苏南京211167

出　　处：《河海大学学报（自然科学版）》2009年第1期91-95,共5页Journal of Hohai University(Natural Sciences)

基　　金：国家自然科学基金(50279008)

摘　　要：为提高岩土材料微结构特征参数提取的准确性和完整性,运用小波变换的多分辨率技术提取并重组了同组图像序列中的清晰部分,进而得到了高清晰度的图像,随后对3种常用的阈值分割方法在岩土材料微结构图像预处理中的适用性进行了对比分析.结果表明:利用小波变换的多分辨率技术重组岩土材料微结构图像,能够有效增大图像信噪比,减小灰度方差,优化微结构数字图像的质量;应用平均灰度法和最大方差自动取阈法处理岩土材料微结构图像,能够有效分割图像并保证特征参数提取的完整性.In order to increase the accuracy and integrality of extracting microstructural characteristic parameters of geomaterials, clear parts of the same image array were extracted and recombined by use of the wavelet multi-resolution technology. Thus the images with high definition were acquired. A comparative analysis for the applicability of 3 frequently used threshold image segmentation methods to process microstructural images of geomaterials was conducted. The results indicate that the recombination of microstructural images of geomaterials by use of the wavelet multi-resolution technology can improve the signal-to-noise ratio of images, reduce the gray scale variance and optimize the microstructural images of geomaterials; the average gray scale method and the automatic threshold extraction method according to the maximum variance to deal with the microstructural images of geomaterials can effectively segment images and ensure the integrality of extracting microstructural characteristic parameters of geomaterials.

关 键 词：岩土材料 微结构 图像融合 图像分割 多分辨率技术 

分 类 号：TU411.92[建筑科学—岩土工程]