[3DSAp2/3Dp2] 3D and Hyper-realistic Displays and Applications 2

[3DSAp2/3Dp2-13] Investigation of Single-Pixel Imaging using Recurrent Neural Network

*Ikuo Hoshi¹, Tomoyoshi Shimobaba¹, Takashi Kakue¹, Tomoyoshi Ito¹ (1. Chiba University (Japan))

キーワード：Single-Pixel Imaging, Deep Learning, Recurrent Neural Network

We propose a reconstruction method for single-pixel imaging. Recently, reconstruction methods using deep neural networks have been studied. However, these methods need much calculation. In this paper, we investigated to reconstruct images from a single-pixel device using a recurrent neural network and decrease the calculation amount.