基于数字图像处理与分析技术,在砂土模型试验中利用模型箱表面拍取的照片获取砂土细观结构参数。首先对试验操作提出了一系列要求,使得可以通过对图像灰度值设立一个适当的阀值来粗略区分砂颗粒与孔隙。然后利用形态学分水岭分割技术,结合图像平滑、区域标记、区域融合等图像预处理与后处理步骤,获取边缘图像,从中提取砂颗粒边缘坐标值及其它相关信息,进而提出一个求解砂土细观结构参数的基本框架,在这个框架下,砂颗粒大小形状的参数、孔隙大小形状的参数。以及砂粒间接触关系的参数都是可求解的。最后对试验条件的局限性进行了讨论,并利用自行开发的软件展示了一个简单的实例。
Based on the digital image processing and analysis technology, the microstructure parameter of granular soils was computed from the photos taken from the surface of the model box in granular soil model tests. Firstly, some requirements were proposed for the operation in test to make sure that granule and hole could be differentiated approximately by a threshold of magnitude of the image. Then, by use of morphologic watershed segmentation as well as the image pretreatment steps such as smoothing, region marking and region merging, the edge image was acquired. The coordinates of granular edge and other necessary data could be gained by scanning this image. Based on the above data, a basic structure was prompted, from which the parameters about the area and shape of granules, the parameters about the area and shape of holes, and the parameters about the contact relation of two granules could all be computed. Finally, the limitation of the method was discussed and a simple example was given.