基于模糊边缘检测的复杂晶界的重建

提出了一种基于模糊边缘检测提取复杂晶界的实用方法。首先通过预处理调整图像对比度、消除晶粒内部灰度差和划痕，进而消除伪边界；再引入模糊理论来判别和跟踪边界，模糊边缘检测算法的特点是不需要确定门限值，具有很强的自适应性；最后进行细化去枝、修补等得到单像素宽度的晶界。实验结果显示，该方法在有效抑制伪边界的同时提取出了比较理想的晶界，为晶粒度分析创造了良好的条件。

COMPLICATED GRAIN-BOUNDARY RECONSTRUCTION BASED ON FUZZY EDGE DETECTION

A practical algorithm based on fuzzy edge detection is proposed. The first step of this algorithm is to adjust the image contrast and eliminate the gray scale difference and nicks inside the grain by using pretreatment. Then we employ the gradient operator and fuzzy theory to judge and trace the edges, which can automatically find gradient thresholds. At last, we get the one pixel wide grain-boundary by thinning, removing pseudo-structures, repairing and so on. The experiments show that the grain-boundary are obtained while the fake edges are restrained, which makes a nicer condition for grain size analysis.