一种基于边缘与分区的棒材端面图像的分割方法

提出了一种利用图像边缘检测与图像分区的综合图像分割的方法。首先运用索贝尔边缘检测技术对原图像副本提取边缘并二值化，然后，对图像分区，结合各分区内边缘点集的特点，分别由全局阈值、子块阈值及分区扩大后阈值对各个分区实施合理分割。实验证明：该分割技术实现简单，应用于具有光照不均匀、子目标粘连、变形等特点的棒材端面图像的分割，效果十分理想。与已有分割方法的对比实验表明，该方法具有目标分割完整，子目标分割也较为合理的特点。

An edge and region-based image segmentation method for the bar section

A comprehensive method of the bar section image segmentation, which used image edge detection and image region, was put forward. First, the copy of the source image whose edges were detected by the Sobel was converted to the binary image. Second, to the image regions, unifying in various regions' edge set of points characteristic, each region implemented the reasonable image segmentation using the overall threshold, the sub- block threshold or the region expands threshold. The experiments proved that, this technology has the virtues of realization simple and the extremely ideal effect when applied to the bar section which have the bad illumination condition, some sub-targets adhesion, distortion and so on. The technology also has the characteristic that the image segmentation is integrity and the sub-image segmentation is more reasonable than the common methods.