带钢表面缺陷图像的增强和分割方法

对带钢表面缺陷进行检测时,由于光照不均匀,将导致缺陷难以识别和分割。针对此问题,提出了改进的图像增强与分割方法。首先,利用自适应二维高斯函数对图像背景进行估计,并结合图像的像素运算均匀图像背景灰度;然后,采用灰度变换函数提高缺陷区域与背景的对比度,增强细节信息;最后,采用最大相关准则方法选取阈值对缺陷图像进行分割。实验结果表明,与典型的图像增强和分割方法相比,所提出的方法表现优异,对非均匀光照下带钢表面缺陷图像的识别与分割具有较好的应用价值。

Enhancement and Segmentation Method of Strip Steel Surface Defect Image

Due to non-uniform illumination, the defect is difficult to recognize and segment when the surface of strip steel is detected. Aiming at these problems, an improved image enhancement and segmentation method is proposed. First, an adaptive two-dimensional Gaussian function is used to estimate the background of the image, and combined with pixel operation to make the image background gray level uniform. Then, a gray transformation function is used to improve the contrast between the defective area and the background, and enhance defect details. Finally, the maximum correlation criterion method is used to select the threshold to segment the defect image. The experimental results show that, compared with the typical image enhancement and segmentation method, the proposed method performs well, and has good application value for the recognition and segmentation of strip steel surface defects under non-uniform illumination.