摘要
深度学习是目前最强大的机器学习算法之一,其中卷积神经网络模型具有自动学习特征的能力,在图像处理领域较其他深度学习模型有较大的性能优势。该文先简述了深度学习的发展史;然后综述了深度学习在超声检测缺陷识别中的应用与发展,从早期浅层神经网络到现在深度学习的应用现状,并借鉴医学影像识别和射线图像识别领域的方法,分析了卷积神经网络对超声图像缺陷识别的适用性;最后,探讨归纳了目前在超声检测图像识别中使用卷积神经网络存在的一些问题及其主要应对策略的研究方向。
Deep learning is one of the most powerful machine learning algorithms and convolutional neural network (CNN) can automatically extract features which outperforms other deep learning model in the eld of image processing. We brie y describe the development history of deep learning, then summarize the application of deep learning in ultrasonic testing defect recognition which from the early shallow neural network to the deep learning. Learning from the eld of medical image recognition and ray image recognition, we nd that the CNN is also suitable for ultrasonic image identi cation, so we propose to use it to identify the ultrasound images directly. Finally, we discuss the problems and practicable strategies in ultrasonic image recognition using CNN.
作者
李萍
宋波
毛捷
廉国选
LI Ping;SONG Bo;MAO Jie;LIAN Guoxuan(State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing 100190, China;University of Chinese Academy of Sciences, Beijing 100049, China)
出处
《应用声学》
CSCD
北大核心
2019年第3期458-464,共7页
Journal of Applied Acoustics
基金
国家杰出青年科学基金项目(11504403)
关键词
深度学习
超声检测
缺陷识别
卷积神经网络
Deep learning
Ultrasonic testing
Flaw recognition
Convolution neural network
