【摘要】 航空材料在工业生产中会不可避免地出现缺陷问题，不及时检测与处理会导致严重的安全问题。传统的人工检测方法耗时长，误检率漏检率高，因而利用深度学习方法对材料进行缺陷的检测越来越被研究者关注。论文提出一种基于局部与全局特征的超声C扫描图像缺陷分割网络。该网络由一条下采样路径和一条上采样路径组成。为了更好地提取图像的全局与局部特征，在下采样路径中加入了注意力机制。实验结果表明，该方法明显优于经典的分割网络，能够快速、准确地对缺陷位置进行分割检测。更多还原

【Abstract】 Aviation materials will appear defects inevitably in industrial production,the untimely detection and treatment will lead to serious safety problems. Traditional manual detection methods are time-consuming and have a high rate of false detection and missed detection. Therefore,deep learning methods are used to detect material defects has attracted more and more researchers’ attention. An ultrasonic C-scan image defect segmentation network based on local and global features is proposed. The network consists of a down-sampling path and an up-sampling path. In order to extract global and local features of images better,attention mechanism is added to the up-sampling path. Experimental results show that this method is better than classical segmentation network and can detect the defect location quickly and accurately.更多还原