基于图像增强的手机触摸屏缺陷检测技术研究
Research on Mobile Phone Touch Screen Defect Detection Based on Image Enhancement
摘要: 在缺陷检测分类的应用中，深度学习的方法逐渐成为主流，变成大家普遍选择的方法，它具有速度快，效率高，准确率高等优点，但同时它也要求具备大量的缺陷数据来用于模型训练，而在实际的场景应用里，我们很难获取到大量的缺陷数据。因此，为增加手机缺陷检测的数据集种类及数量，在此提出一种基于数学参数方程来模拟生成图像数据的方法。利用缺陷的形状特点，将缺陷大致分为点缺陷、块缺陷、线缺陷三种，结合数学方程来模拟缺陷特征信息，然后与无缺陷图像进行融合，最终生成大量的缺陷屏幕的数据。最后通过深度学习的方法来验证此数据集的可用性，通过真实数据和生成数据的分类结果对比发现，使用了生成数据集的模型训练结果提高了1.23%。
Abstract: In the application of defect detection classification, the method of deep learning has gradually become the mainstream and has become the method of general choice, which has the advantages of speed, high efficiency and high accuracy, but at the same time it also requires a large amount of defect data for model training, and in the actual scenario application, it is difficult for us to obtain a large number of defect data. Therefore, in order to increase the type and number of mobile phone defect detection datasets, a method based on mathematical parameter equations to simulate the generation of image data is proposed here, a large number of defect screen data is generated, and finally the usability of this dataset is verified by deep learning, and the model training results using the generated data set are improved by 1.23% by comparing the classification results of real data and generated data.
文章引用：杨淑惠, 于大泳. 基于图像增强的手机触摸屏缺陷检测技术研究[J]. 建模与仿真, 2024, 13(3): 3154-3164. https://doi.org/10.12677/mos.2024.133289

参考文献

[1] Xiang, Y.X., Li, T.T., Ren, W., et al. (2023) A Lightweight Privacy-Preserving Scheme Using Pixel Block Mixing for Facial Image Classification in Deep Learning. Engineering Applications of Artificial Intelligence, 126, Article ID: 107180. [Google Scholar] [CrossRef] 
[2] Kheddar, H., Himeur, Y., Maadeed, S., et al. (2023) Deep Transfer Learning for Automatic Speech Recognition: Towards Better Generalization. Knowledge-Based Systems, 277, Article ID: 110851. [Google Scholar] [CrossRef] 
[3] Wang, R.G., Chen, Z.Y. and Li, W.H. (2023) Gradient Flow-Based Meta Generative Adversarial Network for Data Augmentation in Fault Diagnosis. Applied Soft Computing, 142, Article ID: 110313. [Google Scholar] [CrossRef] 
[4] Li, S.Y. and Zhao, X.F. (2023) High-Resolution Concrete Damage Image Synthesis Using Conditional Generative Adversarial Network. Automation in Construction, 147, Article ID: 104739. [Google Scholar] [CrossRef] 
[5] Siu, C.F., Wang, M.Z., Cheng, J.C.P., et al. (2022) A Framework for Synthetic Image Generation and Augmentation for Improving Automatic Sewer Pipe Defect Detection. Automation in Construction, 137, Article ID: 104213. [Google Scholar] [CrossRef] 
[6] 郭建平. 宏程序在斜椭圆数控车削中的编制方法[J]. 北京工业职业技术学院学报, 2021, 20(3): 6-11.
[7] 张琦, 区锦峰, 周华英. 基于Xception与迁移学习的中药饮片图像识别研究[J]. 现代电子技术, 2024, 47(3): 29-33.
[8] 刘荣光, 朱传军, 成佳闻, 等. 基于改进VGG13的冲压件表面缺陷识别方法研究[J]. 机床与液压, 2024, 52(2): 199-203.
[9] Yang, L., Yu, X.Y., Zhang, S.P., et al. (2023) GoogLeNet Based on Residual Network and Attention Mechanism Identification of Rice Leaf Diseases. Computers and Electronics in Agriculture, 204, Article ID: 107543. [Google Scholar] [CrossRef] 
[10] Razavi, M., Mavaddati, S. and Koohi, H. (2024) ResNet Deep Models and Transfer Learning Technique for Classification and Quality Detection of Rice Cultivars. Expert Systems with Applications, 247, Article ID: 123276. [Google Scholar] [CrossRef] 
[11] Li, B. and Lima, D. (2021) Facial Expression Recognition via ResNet-50. International Journal of Cognitive Computing in Engineering, 2, 57-64. [Google Scholar] [CrossRef] 

为你推荐