Spectrum classification using convolutional neural networks for a mini-camera detection system
| Autor: | Haiyang Zhang, Chun Liu, Changming Zhao, Zilong Zhang, Zitao Cai, Zhipeng Li |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
medicine.medical_specialty
Artificial neural network Computer science business.industry ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Image processing Speckle noise 01 natural sciences Convolutional neural network Atomic and Molecular Physics and Optics Spectral imaging 010309 optics Optics 0103 physical sciences medicine Computer vision Artificial intelligence Electrical and Electronic Engineering business Engineering (miscellaneous) |
| Zdroj: | Applied Optics. 58:9230 |
| ISSN: | 2155-3165 1559-128X |
| Popis: | A mini-camera is one of several emerging cat-eye devices featuring tiny lenses with diffracted retro-reflections. It is hard for traditional active laser detection systems to identify a mini-camera because of their weak reflection. This paper proposes an anti-camera system with a spectrum-based convolutional neural network algorithm to recognize the profile features of the retro-reflection images captured by the system. The network was trained with the spatial spectra of local datasets and uploaded onto the embedded device. The results of several indoor experiments demonstrate that the system reached high accuracy in real-time detection, even with various types of interference. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|