Spectrum classification using convolutional neural networks for a mini-camera detection system.

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.

[1]  Yongli Zhao,et al.  SOON: self-optimizing optical networks with machine learning. , 2018, Optics express.

[2]  Mali Gong,et al.  Periodicity analysis on cat-eye reflected beam profiles of optical detectors , 2017 .

[3]  Lixin Zheng,et al.  Combining Convolutional Neural Network With Recursive Neural Network for Blood Cell Image Classification , 2018, IEEE Access.

[4]  J. Nathan Kutz,et al.  Classification of birefringence in mode-locked fiber lasers using machine learning and sparse representation. , 2014, Optics express.

[5]  Nitin Khanna,et al.  DCT-domain Deep Convolutional Neural Networks for Multiple JPEG Compression Classification , 2017, Signal Process. Image Commun..

[6]  Lambert Schomaker,et al.  DeepOtsu: Document Enhancement and Binarization using Iterative Deep Learning , 2019, Pattern Recognit..

[7]  Jee-Young Sun,et al.  A novel contrast enhancement forensics based on convolutional neural networks , 2018, Signal Process. Image Commun..

[8]  Michael Unser,et al.  Deep Convolutional Neural Network for Inverse Problems in Imaging , 2016, IEEE Transactions on Image Processing.

[9]  Li Li,et al.  Recognizing cat-eye targets with dual criterions of shape and modulation frequency , 2011 .

[10]  M Soriano,et al.  Fluorescent image classification by major color histograms and a neural network. , 2001, Optics express.

[11]  Bo Liu,et al.  Compressive sensing method for recognizing cat-eye effect targets. , 2013, Applied optics.

[12]  Shuai Li,et al.  Spectral pre-modulation of training examples enhances the spatial resolution of the phase extraction neural network (PhENN). , 2018, Optics express.

[13]  Mali Gong,et al.  Cat-eye effect reflected beam profiles of an optical system with sensor array. , 2016, Applied optics.

[14]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.