Adaptive-sampling angular spectrum method with full utilization of space-bandwidth product.

As convolution-based diffraction calculation methods, Rayleigh-Sommerfeld convolution and the angular spectrum method (ASM) usually require zero padding to avoid circular convolution errors. This greatly increases the computational complexity and wastes a large amount of the sampling points. In this Letter, based on the analysis of sampling properties in the convolution process, we propose an adaptive-sampling ASM, which can adjust the sampling parameters according to the propagation distance to avoid circular convolution errors without zero padding. The sampling condition of the transfer function can be adaptively satisfied by rearranging the sampling points in the spatial frequency domain. Therefore, the computational complexity is significantly reduced, and all the sampling points are effectively used, which leads to a full utilization of the space-bandwidth product.