利用扫描电镜图像中边缘像素的坡向与颗粒的长轴定向方向正交的原理,采用GIS空间分析技术对扫描电镜图像进行边缘提取和定向分析,并分别利用定向玫瑰图、主定向方向和各向异性率等指标对颗粒的定向特征进行描述,通过实例分析验证该方法的有效性。试验结果表明:由于噪声及算法误差的影响,低坡度区间内的定向玫瑰图及各向异性率不能反映土样的定向特征,玫瑰图在坐标轴及对角线方向上将出现异常突起,各向异性率也明显偏低;随着坡度的升高,噪声及算法误差的影响逐渐变小,并可以很好地反映土样的整体定向特征。
A new technique for quantitative analysis of the orientation distribution of soil grains from SEM images, slope-aspect technique, was described in this paper. The pixels at the edge of the particles of SEM were abstracted by the method of slope analysis. Because the aspect values of the pixels at the edge of the particles are orthogonal to the orientation of the particles, the orientation characteristics of a SEM image of soil samples can be obtained by quantitatively analyzing the aspect of the pixels at the edge of the particles in SEM. Some quantitative analysis techniques, including rosette diagrams, index of anisotropy and main orientation angle were also used. Through some examples, the slope-aspect technique is proved to be an efficient technique in examining orientation patterns of clay or sand soil microstructure, and it can provide more reasonable rosette histogram, index of anisotropy and main orientation angle. It also shows that the results in low slop classes can not present the orientation characteristics of whole sample because of noise and algorithmic error. But aspect values of pixels with higher slope class on the boundaries of particles can well present the orientation pattern of soil samples.