torchsar.calibration package¶

Submodules¶

torchsar.calibration.channel_process.iq_correct(Sr)¶

IQ Correction performs the I Q data correction

see “Sentinel-1-Level-1-Detailed-Algorithm-Definition”.

Parameters: Sr (Tensor) – SAR raw data matrix \({\bm S}_r \in {\mathbb R}^{N_a×N_r×2}\)
Returns: Corrected SAR raw data. Flag : dict
Return type: Sr (Tensor)

torchsar.calibration.gain_compensation.vga_gain_compensation(S, V, mod='linear', fact=1.0)¶

vga gain compensation

\[\begin{aligned} {\bm F} &= (λ 10^{{\bm V}/20})\\ {\bm S}_{c} &= {\bm F} \odot {\bm S} \end{aligned} \]

Parameters

S (torch tensor) – S is an \(N_a×N_r×2\) array, where, \(S[:,:,0]\) is the I signal and \(S[:,:,1]\) is the Q signal.
V (torch tensor) – S is an \(N_a×N_r\) or \(N_a×1\) VGA gain array, the gain values are in dB unit.
mod (str, optional) – compensation mode (the default is ‘linear’)
fact (number, optional) – fact is the factor \(\lambda\) (the default is 1.0)

Returns

compensated signal, \(N_a×N_r×2\) array.

Return type

torch tensor

torchsar.calibration.multilook_process.multilook_spatial(Sslc, nlooks)¶

spatial multilook processing

spatial averaging in azimuth or range direction.

Parameters

Sslc (Tensor) – Processed single look complex (or intensity) sar data tensor with size \(N_a×N_r\).
nlooks (tuple or list) – The number of looks in azimuth and range direction, [na, nr] or (na, nr).

Returns

Processed multi-look complex tensor.

Return type

Smlc (Tensor)