中文摘要：

提出由后续结果衡量图像预处理效果的评价标准，并根据Radon变换的结果对中值滤波算法进行反馈校正，改进后的算法能自动确定滤波模板尺寸和滤波次数．应用中值滤波预处理与Radon变换结合的方法对图像标定符号进行定位克服了强噪声背景下单独采用连通域定位方法的缺陷．实验证明，改进后的算法增强了金属断口图像的纹理特征，对在图像中所占比例较小的标尺符号定位更加准确，并可根据标尺符号的位置、长度信息以及标定字符的纹理特征对强噪声背景下的标定符号进行准确分割．

英文摘要：

Proposed the standard of evaluating the effect of image enhancement according to the results of subsequent proceeding. Median filter is feedback corrected by the results of Radon transform so that the size of template and times of median filter are determined automatically. The algorithm of improved median filter based on Radon transform avoids the deficiency of localization with connectivity regions solely in complex background. The experiment results show that the improved algorithm enhances texture feature of metal fracture image and is more effective in the localization of calibrated ruler that has a less proportion of the whole image. Furthermore, it accurately segments calibrated characters from strong noisy background based on texture feature of characters as well as the position and length of ruler symbol.