원문정보
보안공학연구지원센터(IJSIA)
International Journal of Security and Its Applications
Vol.10 No.6
2016.06
pp.269-280
피인용수 : 0건 (자료제공 : 네이버학술정보)
초록
영어
There are shortcomings in the currently used traditional CCD imaging system which can automatically recognize banknote image defect, such as the need to manually extract the defect characteristics and low accuracy rate of detection results. This paper briefly introduced the advantage of convolution Neural Network (CNN) in image classification and designed a image defect identification method based on convolutional neural network (CNN). The experimental results on data sets show that the identification accuracy rate of this method is 95.6%, which is significantly better than traditional identification method.
목차
Abstract
1. Introduction
2. Convolution Neural Network
2.1. Convolution Neural Network Structure
2.2. Convolution Layer
2.3. Low Sample Layer
2.3. Pooling
2.4. All-Connection Layer
2.5. Training Method
3. Banknote Image Defect Recognition Method based on Convolution Neural Network
3.1. C1 Layer
3.2. S1 Layer
3.3. C2 Layer
3.4. Other Convolution Layers and Sub-sampling Layers
3.5. Output Layer
4. Experiment Results and Analysis
4.1. Dataset Collection
4.2. Experiment Results Analysis
5. Conclusion
Acknowledgements
References
1. Introduction
2. Convolution Neural Network
2.1. Convolution Neural Network Structure
2.2. Convolution Layer
2.3. Low Sample Layer
2.3. Pooling
2.4. All-Connection Layer
2.5. Training Method
3. Banknote Image Defect Recognition Method based on Convolution Neural Network
3.1. C1 Layer
3.2. S1 Layer
3.3. C2 Layer
3.4. Other Convolution Layers and Sub-sampling Layers
3.5. Output Layer
4. Experiment Results and Analysis
4.1. Dataset Collection
4.2. Experiment Results Analysis
5. Conclusion
Acknowledgements
References
저자정보
참고문헌
자료제공 : 네이버학술정보