Method and system for sar imaging using complexvalued local interpolation function based range migration algorithm

A system, comprising:

• one or more hardware processors;
• a communication interface;
• and
• a memory storing a plurality of instructions which cause the one or more hardware processors to:
• receive a spatially irregular sampled complex raw data from a radar, as input data, wherein the spatially irregular sampled complex raw data is captured by the radar while moving along a trajectory with an irregular scanning geometry;
• transform the spatially irregular sampled complex raw data into a two-dimensional (2D) feature map, using a complex-valued encoder of a trained complex-valued Local Interpolation Function (LIF) network;
• predict an interpolated complex raw data at one or more missing data points at one or more spatial coordinates, using a plurality of nearest latent codes from the 2D feature map and a plurality of query coordinates corresponding to a plurality of missing data points, by using complex-valued decoder of the trained complex-valued LIF network, wherein the plurality of nearest latent codes mitigate distortions caused by a plurality of irregular sampling and interpolation errors, wherein the predicted interpolated complex raw data is computed using a weighted average of the plurality of nearest latent codes, with associated plurality of weights determined by a Euclidean distance between each of the plurality of query coordinates (x, y) and each of the plurality of nearest latent codes;
• transform the interpolated complex raw data into a frequency domain by using a 2D Fast Fourier Transform;
• and
• generate a reconstructed image based on the transformed frequency domain of interpolated complex raw data.

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:

• receiving a spatially irregular sampled complex raw data from a radar, as input data, wherein the spatially irregular sampled complex raw data is captured by the radar while moving along a trajectory with an irregular scanning geometry;
• transforming the spatially irregular sampled complex raw data into a two-dimensional (2D) feature map, using a complex-valued encoder of a trained complex-valued Local Interpolation Function (LIF) network;
• predicting an interpolated complex raw data at one or more missing data points at one or more spatial coordinates, using a plurality of nearest latent codes from the 2D feature map and a plurality of query coordinates corresponding to a plurality of missing data points, by using complex-valued decoder of the trained complex-valued LIF network, wherein the plurality of nearest latent codes mitigate distortions caused by a plurality of irregular sampling and interpolation errors, wherein the predicted interpolated complex raw data is computed using a weighted average of the plurality of nearest latent codes, with associated plurality of weights determined by a Euclidean distance between each of the plurality of query coordinates (x, y) and each of the plurality of nearest latent codes;
• transforming the interpolated complex raw data into a frequency domain by using a 2D Fast Fourier Transform;
• and
• generating a reconstructed image based on the transformed frequency domain of interpolated complex raw data.