Abstract

Many existing polarization networks reconstruct polarization information based on calculating the angle of polarization (AoP) loss. Yet, the conventional loss calculation method, which is based on a linear difference approach, compromises the reconstruction accuracy and causes additional training time when combined with learning-based methods. In this Letter, we present a new, to the best of our knowledge, method to calculate the AoP loss and apply it in an enhanced color polarization demosaicking network with a “multi-branch” structure, i.e., ePDNet. Experiments are performed to demonstrate the efficacy and superiority of the method, which improves the network convergence speed by three times as well as the output image quality. The new method may find important applications in the field of polarimetric imaging.

© 2022 Optica Publishing Group

Color polarization demosaicking by a convolutional neural network

Yuanyuan Sun, Junchao Zhang, and Rongguang Liang
Opt. Lett. 46(17) 4338-4341 (2021)

Unsupervised demosaicking network using the recurrent renovation and the pixel-wise guidance

Jinyang Li, Jia Hao, Geng Tong, Shahid Karim, Xu Sun, and Yiting Yu
Opt. Lett. 47(16) 4008-4011 (2022)

Attention-based neural network for polarimetric image denoising

Hedong Liu, Yizhu Zhang, Zhenzhou Cheng, Jingsheng Zhai, and Haofeng Hu
Opt. Lett. 47(11) 2726-2729 (2022)

Convolutional demosaicing network for joint chromatic and polarimetric imagery

Sijia Wen, Yinqiang Zheng, Feng Lu, and Qinping Zhao
Opt. Lett. 44(22) 5646-5649 (2019)

IPLNet: a neural network for intensity-polarization imaging in low light

Haofeng Hu, Yang Lin, Xiaobo Li, Pengfei Qi, and Tiegen Liu
Opt. Lett. 45(22) 6162-6165 (2020)