中文摘要：

图像预处理是数字图像处理领域的重要部分，也是冷轧带钢表面缺陷图像检测的前提。复杂的现场环境及光学系统的失真等因素可导致图像降质，进而直接影响后续图像的特征提取及分类。针对这些问题，采取自适应中值滤波与同态滤波相结合的方法对图像进行预处理，利用自适应中值滤波实现图像去噪，高斯同态滤波较稳定地去除图像不均勾问题；将未经预处理与经过预处理后的图像及图像特征进行分析与比较，结果表明，微处理较大地改善了图像质量。

英文摘要：

The image preprocessing is an important part in the field of digital image processing, and it?s also the premise for the image detection of cold steel strip surface defects. The factors including the complicated on-site environment and the distortion of the optical system will cause image degradation, which will directly affects the feature extraction and classification of the images. Aiming at these problems, a method combining the adaptive median filter and homomorphic filter is proposed to preprocess the image. The adaptive median filter is effective for image denoising, and the Gaussian homomorphic filter can steadily remove the nonuniform illumination of images. Finally, the original and preprocessed images and their features are analyzed and compared. The results show that this method can improve the image quality effectively.