关键词: 岩石图像, 多尺度分割, 图像处理, 矿物特征提取

Abstract:

Traditionally, minerals identification of thin sections under the microscope is carried out with visual interpretation and qualitative description. It is subjective and difficult to quantify.We introduce a digital image processing technique to minerals analysis, and multiresolution segmentation is employed to mineral features extraction and analysis on the petrographic image of quartz sandstone. Firstly, we obtained the boundary of the quartz grains using multiresolution segmentation. Then the features of quartz particles were extracted, including size, contents, perimeter, long axis length and orientation, the long axis and minor axis ratio，roundness and shape index; pixel statistics shows quartz content are of 47.07%, less than the matrix, which floating in the matrix with matrix support structure; the long axis of quartz ranges from 50 to 604 pixels, which shows the size of quartz grains is poorly sorted; and the roundness is 0.27-1.82, shape index value of 1.19-2.46, roundness and shape index is of small value, that means it has certain psephicity. This study shows digital image processing based on multiresolution segmentation can achieve the quantitative description for the minerals features, and it is helpful for petrographic images understanding and analysis.

Key words: petrographic image, multiresolution segmentation, image processing, mineral features extraction