Using freeform optical surfaces is a revolution in the field of imaging system design. Such systems have important applications in the area of virtual reality and augmented reality, light-field and high-performance cameras, microscopy, spectroscopy, and other applied physics researches. We propose a framework of starting points generation for freeform reflective imaging systems using back-propagation (BP) neural network based deep-learning. Good starting points of specific system specifications for optimization can be generated immediately using the network. The amount of time and human effort as well as the dependence on advanced design skills reduce significantly.

© 2019 Japan Society of Applied Physics, The Optical Society (OSA)

PDF Article
More Like This
Deep Learning-designed Diffractive Neural Networks

Xing Lin, Yair Riverson, Nezih T. Yardimci, Muhammed Veli, Yi Luo, Mona Jarrahi, and Aydogan Ozcan
SF1N.2 CLEO: Science and Innovations (CLEO_SI) 2019

Learning Biological Structures from Birefringence images with Deep Neural Networks

Syuan-Ming Guo, Anitha Krishnan, Jenny Folkesson, Jim Karkanias, and Shalin B. Mehta
MTu3D.3 Mathematics in Imaging (MATH) 2019

Construction-iteration method for the design of freeform reflection system

Benqi Zhang, Guofan Jin, and Jun Zhu
FM3B.1 Freeform Optics (Freeform) 2019


You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
Login to access OSA Member Subscription