基于图像边缘信息和Fisher准则的钢板表面缺陷分割研究

Study on defection segmentation for steel surface image based on image edge detection and Fisher discriminant

针对钢板表面缺陷图像信噪比低、缺陷目标小且形态差别大等特点,提出了一种基于边缘信息和Fisher准则相结合的图像分割方法。该方法首先采用梯度算子检测出缺陷图像的边缘,并对边缘检测所得的梯度图进行灰度拉伸,提高梯度图的对比度;然后利用Fisher准则寻找最佳阈值,分割出缺陷;最后运用数学形态学滤除噪声,实现了缺陷的自动分割和定位。实验证明,该方法不仅能够识别出弱小的缺陷,而且实现了在线实时检测。

A hybrid image segmentation method based on edge detection and Fisher discrirninant was presented to detect defection, because signal-to-noise ratio of steel surface image is very low, and defection targets are small and their shape is irregular. Firstly, gradient operator detect the edge of defection image and gradient image was gotten, then grayscale of gradient image was stretched in order to enhance image contrast. Secondly, Fisher discriminant was adopted in order to find optimum threshold, meanwhile defection targets were segmented. Lastly, noise was filtered by morphology method. Defection was auto- segmented and located by this segmentation method. Experiment results show this method can detect week defection and realtime detect defection online.