摘要
缺陷检测是图像处理工作的重要步骤,其检测效果直接影响图像的后期修复和调节。由于目前基础检测系统对图像缺陷的识别效果较差,因此设计了基于深度学习的图像缺陷检测系统。系统硬件设计,构建分级化处理单元模块,并优化卷积神经网络(Convolutional Neural Networks,CNN)双向电路。系统软件设计,设计缺陷分割层级模块,建立缺陷特征检测数据库。结果表明,得出的预测识别精度均控制在90%以上,系统在实际应用过程中对图像缺陷的识别速度较快,同时检测误差较小,效果更佳,具有实际的应用价值。
Defect detection is an important step in image processing.The effect of defect detection directly affects the post-repair and adjustment of images.Because the current basic detection system has poor recognition effect on image defects,an image defect detection system based on deep learning is designed.System hardware design,build hierarchical processing unit module,and optimize convolutional neural networks(CNN)bidirectional circuit.System software design,design defect segmentation level module,and establish defect feature detection database.The results show that the predicted recognition accuracy is controlled above 90%,which shows that the system can recognize image defects faster in the actual application process,and the detection error is smaller,and the effect is better,which has practical application value.
作者
秦丽娜
QIN Lina(Information Engineering College,Zhengzhou Shengda University,Zhengzhou Henan 450000,China)
出处
《信息与电脑》
2022年第16期91-93,共3页
Information & Computer
基金
河南省科技厅2022年度科技发展计划项目“智慧城市政务智能一体化关键技术研究”(项目编号:222102210290)。
