Abstract
Centroid-Predicted Deep Neural Network in Shack-Hartmann Sensors
Highlights
THE Shack-Hartmann wavefront sensor (SHWFS) has been widely used in astronomy [1], biological imaging [2, 3], ophthalmology [4, 5], and laser beam characterization [6, 7]
All the above methods demonstrate that machine learning can improve the performance of the SHWFS
The remaining part of the paper is organized as follows: Section II.A introduces the measurement principle of the SHWFS and the measurement error introduced due to the absence of sub-spots, Section II.B describes the structure of the constructed deep neural network, and Section II.C introduces the method of generating datasets; In Section III, the numerical simulation results are displayed to demonstrate the feasibility of the proposed model
Summary
THE Shack-Hartmann wavefront sensor (SHWFS) has been widely used in astronomy [1], biological imaging [2, 3], ophthalmology [4, 5], and laser beam characterization [6, 7]. We first treat the inaccurately detectable spots as missing spots and work on retrieving their information through the machine learning to improve the measurement precision of the SHWFS. We use a deep neural network to predict the missing sub-spots’ position from the centroid-displacement information of detected sub-spots. This method enables the SHWFS more tolerant to the absence of some sub-spots. The remaining part of the paper is organized as follows: Section II.A introduces the measurement principle of the SHWFS and the measurement error introduced due to the absence of sub-spots, Section II.B describes the structure of the constructed deep neural network, and Section II.C introduces the method of generating datasets; In Section III, the numerical simulation results are displayed to demonstrate the feasibility of the proposed model.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have