Curvelet结合图像方向离散度的DSPI无损检测缺陷边缘提取

Curvelet combined orientation dispersion of image in defect edge detection of DSPI nondestructive testing

边缘提取是DSPI无损检测缺陷自动识别的基础。根据人类视觉识别的特点,提出以Curvelet多尺度几何分析为基础进行缺陷图像的边缘提取,为了对DSPI检测过程中因光照不足等原因产生的伪缺陷进行分辨,基于矢量积定义了图像的方向离散度,并与Curvelet多尺度分析方法相结合,在橡胶复合材料DSPI无损检测中进行应用。

Edge detection is the basis of automatic recognition of flaws in DSPI nondestructive testing. According to the characteristics of the human visual recognition, the edge of defect is extracted based on the Curvelet multiscale geo metric analysis. In order to distinguish the pseudo-flaw caused by insufficient light during DSPI testing, the orientation dispersion of image is defined based on vector product. In the end, the orientation dispersion of image combined with Cur- velet is used in DSPI non-destructive testing of rubber composite.