织物悬垂性数字图像轮廓识别的径向扫描算法

Radial scan algorithm for fabric drape digital image contour detection

根据织物悬垂性数字图像轮廓自动识别的需要,提出了一种简单实用的径向扫描算法。首先识别出试样圆盘圆心,并从原始图像中裁剪出有效区域,进行必要的预处理后再用径向扫描算法得到候选轮廓点集,将不在轮廓上的候选轮廓点去掉,顺次连接轮廓点序列,当轮廓点数量足够多时,就可以用轮廓点顺次连接线逼近织物的边缘轮廓曲线。与传统的梯度算子法相比,这种算法有更好的适应性和可靠性,结合实例介绍了该算法的设计和实现过程及特点。

According to the actual requirements of automatic processing of fabric drape images, a simple but practical radial scan algorithm is proposed. First, identify the center of the disk on the sample, cut the valid area from the original image, and conduct the necessary preprocessing, and then, use the radial scan algorithm to find out the preliminary contour points. By removing the points which are not on the contour and connecting the points remained with lines in sequence, a curve can he formed. When the contour points are adequate enough, the curve can be thought as the contour of the sample. Compared with the traditional gradient operator algorithm, the radial scan algorithm is more applicable and reliable. The design, execution process, and characteristics of the radial scan algorithm are expounded with examples.