基于CT图像的混凝土细观结构边缘检测技术

Edge detection of concrete mesostructure with CT image

根据CT图像成像的特征以及图像边缘检测的原理,采用Laplacian方法、Canny方法、Roberts方法、Pre-witt方法、Sobel方法等多种方法分别对CT图像进行了边缘检测.从边缘检测的效果来看,采用Laplacian方法、Canny方法在边缘连续性和完整性上要优于Roberts方法、Prewitt方法、Sobel方法.将Canny方法检测到的边缘进行修正,效果较为理想,在此基础上,通过矢量化处理后导入到有限元软件中,得到了混凝土真实细观结构的有限元网格.研究表明,采用这种方法能有效地提升数值混凝土有限元计算模型的真实性,从而使数值计算成果与物理力学试验成果相吻合.

According to the character of CT image and the theory of edge detection, many methods such as Laplace method, Canny method, Roberts method, Prewitt method and Sobel method, have been used for the edge detection of CT image. The results show that the I.aplace and Canny methods are better than Roberts, Prewitt and Sobel methods on the continuity and intergrality of the edge. An dealer result has been obtained after modified the edge, and then loads into the software of finite element method to gain the concrete mesostructure model. The research results show that the reality of the numerical con crete finite element model could be improved by the method; and the numerical calculation results are in concordance with the mechanics experiment results.