基于Canny算子和轮廓方向图的异形纤维轮廓增强算法

显微包埋采集纤维图像过程中,由于背景光照不均、景深和聚焦等造成纤维图像对比度较低、目标与背景区分不明显的问题.传统的Canny算子对于异形纤维的轮廓提取存在边缘不连续、虚假边缘等不足,尤其是对于粘连纤维无法确定其轮廓.针对这些不足,提出了一种利用轮廓跟踪算法剔除Canny算子产生的噪声边缘,并在方向图的基础上利用矩形方框确定真正的轮廓区域,然后对此区域内的像素进行模糊域非线性拉伸,从而得到增强后的异形纤维图像.对增强后的异形纤维图像进行Otsu二值化处理,其结果表明该算法能够得到准确完整的异形纤维轮廓.

Profiled Fiber Contour Enhancement Algorithm Based on Canny Operator and Profile Directed Graph

During the fiber image acquisition process by microscopic embedding techniques,because of the low fiber image contrast and the fuzzy distinction of target and background which are caused by the uneven illumination background,the profile extrusion,and the fiber breakage,there are many defects in traditional Canny operator profiled fiber contour extraction,such as the discontinuous edge,the false edge and so on.And especially,it can not determine the profile of fiber adhesion.Focusing on these defects,a profi...