基于空域滤波和空间相关的钢丝绳图像分割算法研究

Research of segmentation algorithms of steel wire rope image based on the spatial filtering and the spatial correlation

在钢丝绳断丝实时检测中,由于复杂环境因素的影响和钢丝绳自身结构特点的原因,钢丝绳图像经常受不均匀光照、阴影、油泥、粉末等干扰,导致图像降质严重,给钢丝绳的分割带来困难.本文提出了一种有效、快速的分割算法:首先采用改进的空域同态滤波器对钢丝绳图像进行滤波,去除不均匀光照和削除阴影边缘,然后根据钢丝绳的几何对称性,结合边缘方向信息测度的差异和钢丝绳区域的局部特征统计,消除干扰,实现钢丝绳的分割.实验结果表明,该分割算法能够快速、完整地分割出钢丝绳图像.

In the real - time inspection of broken wires in steel rope, because of the complex environment and its structure feature, the interferences such as non - uniform illumination or shadow edge or oil sludge or power would affect image quality frequently, it is a difficult task to divide the degraded image. In this paper, an effective and fast segmentation algorithm is proposed. Firstly, eliminate the non - uniform illumination and the shadow edges by adaptive homomorphic filtering on the image. Secondly, eliminate the interference according to the geometry symmetry, combined with the difference of the edge orientation information measurement and local characteristics of statistics, and image segmentation is realized. Experiment results show that this segmentation algorithms could divide the steel wire ropes rapidly and integrallty.