pub struct SpectralImagePlaneModel { /* private fields */ }Expand description
Algorithm-facing narrowband spectral model, containing no physical geometry.
All channels share detector shape and object-plane pixel pitch. Each keeps
its own crops, subpixel offsets, pupil, and wavelength. For detector row f,
prediction is gain[f] * sum(weight * local_intensity) + background[f].
Local source powers, source weights, and illumination-frame gains are
included in local_intensity. Detector calibration is applied once.
Implementations§
Source§impl SpectralImagePlaneModel
impl SpectralImagePlaneModel
Sourcepub fn from_experiment(
channels: &[SpectralChannel],
geometry: &SpectralGeometry,
acquisition: SpectralAcquisitionPlan,
image_shape: (usize, usize),
reconstruction_shape: ReconstructionShape,
object_coupling: ObjectCoupling,
) -> Result<Self>
pub fn from_experiment( channels: &[SpectralChannel], geometry: &SpectralGeometry, acquisition: SpectralAcquisitionPlan, image_shape: (usize, usize), reconstruction_shape: ReconstructionShape, object_coupling: ObjectCoupling, ) -> Result<Self>
Resolves every channel atomically and selects a grid from the union of crop bounds.
IDs and wavelengths must be unique, wavelengths finite and positive, and all detector pitches equal. Shared direct k-vectors and configurations requiring spatial registration or resampling are rejected. An explicit reconstruction shape must contain all channels’ interpolation stencils.
§Example
use fpm_rs::{
algorithms::{SpectralAlternatingProjection, SpectralReconstructionAlgorithm},
experiment::{AcquisitionPlan, DirectionList, Optics, SourceCalibration,
SpectralAcquisitionPlan, SpectralChannel, SpectralGeometry},
measurements::MeasurementStack,
model::{ObjectCoupling, ReconstructionShape, SpectralImagePlaneModel},
reconstruction::SpectralReconstructionProblem,
};
let channels: Vec<_> = [450e-9, 630e-9].into_iter().enumerate().map(|(index, wavelength)| {
Ok(SpectralChannel {
channel_id: format!("channel-{index}"),
optics: Optics {
wavelength_vacuum_m: wavelength, objective_na: 0.1,
magnification: 4.0, camera_pixel_size: 6.5e-6,
illumination_refractive_index: 1.0,
objective_medium_refractive_index: 1.0,
defocus_distance: None, pupil_aberration: None,
},
calibration: SourceCalibration::unity(),
acquisition: AcquisitionPlan::all_sources(1)?,
})
}).collect::<fpm_rs::Result<_>>()?;
let geometry = SpectralGeometry::Shared(
DirectionList::from_direction_cosines(vec![[0.0, 0.0]])?.into(),
);
let model = SpectralImagePlaneModel::from_experiment(
&channels, &geometry, SpectralAcquisitionPlan::separate(&[1, 1])?,
(4, 4), ReconstructionShape::Smooth, ObjectCoupling::Independent,
)?;
let measurements = MeasurementStack::from_vec(vec![1.0; 2 * 4 * 4], (4, 4), vec![])?;
let problem = SpectralReconstructionProblem::new(measurements, model)?;
let result = SpectralAlternatingProjection::default().iterations(2).run(&problem)?;
assert_eq!(result.channels.len(), 2);
let continued = fpm_rs::reconstruction::SpectralRunner::new(
SpectralAlternatingProjection::default().iterations(3),
).with_checkpoint(result.checkpoint.unwrap()).run(&problem)?;
assert_eq!(continued.runtime.completed_iterations, 3);Sourcepub fn from_compiled_channels(
channels: Vec<SpectralModelChannel>,
acquisition: SpectralAcquisitionPlan,
object_coupling: ObjectCoupling,
) -> Result<Self>
pub fn from_compiled_channels( channels: Vec<SpectralModelChannel>, acquisition: SpectralAcquisitionPlan, object_coupling: ObjectCoupling, ) -> Result<Self>
Assembles validated wavelength kernels already compiled on one common grid. Used by external dataset converters; rejects missing wavelength metadata, duplicate IDs/wavelengths, incompatible sampling, and uncovered local frames.
Sourcepub fn mixing_diagnostics(
&self,
relative_tolerance: Option<f64>,
) -> Result<SpectralMixingDiagnostics>
pub fn mixing_diagnostics( &self, relative_tolerance: Option<f64>, ) -> Result<SpectralMixingDiagnostics>
Reports the collapsed detector-exposure/channel mode-weight matrix. Each entry sums detector gain × spectral weight × local frame gain × local incoherent source-weight sum. Background and measurement masks/weights are excluded. Different crops and pupils remain separate forward operators; matrix rank alone does not establish nonlinear reconstruction identifiability.
relative_tolerance must be finite and nonnegative; omission uses
max(exposures, channels) * f64::EPSILON. Rank counts singular values
strictly above largest * relative_tolerance. Singular values use a scaled
cyclic one-sided Jacobi SVD, transposing wide matrices. Nonconvergence fails.
§References
NumPy, matrix_rank documentation, for the default numerical-rank threshold. LAPACK 3.12.1, DGESVJ documentation, for one-sided Jacobi SVD; this crate implements its own cyclic rotations.
Sourcepub fn channels(&self) -> &[SpectralModelChannel]
pub fn channels(&self) -> &[SpectralModelChannel]
Borrows the ordered channel metadata and ordinary numerical kernels.
Sourcepub fn acquisition(&self) -> &SpectralAcquisitionPlan
pub fn acquisition(&self) -> &SpectralAcquisitionPlan
Borrows the canonical sparse physical detector plan.
Sourcepub fn object_coupling(&self) -> ObjectCoupling
pub fn object_coupling(&self) -> ObjectCoupling
Returns the explicitly chosen complex-object sharing rule.
Sourcepub fn object_count(&self) -> usize
pub fn object_count(&self) -> usize
Returns the number of independently stored object spectra (one when shared).
Sourcepub fn object_index(&self, channel: usize) -> Result<usize>
pub fn object_index(&self, channel: usize) -> Result<usize>
Returns the stored object index used by a given channel.
Sourcepub fn image_shape(&self) -> (usize, usize)
pub fn image_shape(&self) -> (usize, usize)
Returns the common low-resolution detector (height, width).
Sourcepub fn reconstruction_shape(&self) -> (usize, usize)
pub fn reconstruction_shape(&self) -> (usize, usize)
Returns the common high-resolution object (height, width).
Sourcepub fn frame_count(&self) -> usize
pub fn frame_count(&self) -> usize
Returns the physical detector frame count.
Sourcepub fn validate(&self) -> Result<()>
pub fn validate(&self) -> Result<()>
Validates common sampling, unique IDs/wavelengths, references, and local coverage.
Sourcepub fn forward_intensity(
&self,
spectra: &[ArrayView2<'_, Complex64>],
frame: usize,
) -> Result<Array2<f64>>
pub fn forward_intensity( &self, spectra: &[ArrayView2<'_, Complex64>], frame: usize, ) -> Result<Array2<f64>>
Evaluates a detector frame by calling each canonical single-channel forward model.
spectra contains centered, standard-layout common-grid spectra in stored
object order, with the core’s 1/(height * width) forward FFT normalization:
one per channel for independent objects, one for a shared
object. Returns an owned standard-layout intensity image. Geometry, pupil
sampling, signs, and subpixel interpolation use the ordinary kernels.
Trait Implementations§
Source§impl Clone for SpectralImagePlaneModel
impl Clone for SpectralImagePlaneModel
Source§fn clone(&self) -> SpectralImagePlaneModel
fn clone(&self) -> SpectralImagePlaneModel
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source. Read moreSource§impl Debug for SpectralImagePlaneModel
impl Debug for SpectralImagePlaneModel
Source§impl<'de> Deserialize<'de> for SpectralImagePlaneModel
impl<'de> Deserialize<'de> for SpectralImagePlaneModel
Source§fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error>
fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error>
Auto Trait Implementations§
impl Freeze for SpectralImagePlaneModel
impl RefUnwindSafe for SpectralImagePlaneModel
impl Send for SpectralImagePlaneModel
impl Sync for SpectralImagePlaneModel
impl Unpin for SpectralImagePlaneModel
impl UnsafeUnpin for SpectralImagePlaneModel
impl UnwindSafe for SpectralImagePlaneModel
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Source§impl<T> CloneToUninit for Twhere
T: Clone,
impl<T> CloneToUninit for Twhere
T: Clone,
Source§impl<T> IntoEither for T
impl<T> IntoEither for T
Source§fn into_either(self, into_left: bool) -> Either<Self, Self> ⓘ
fn into_either(self, into_left: bool) -> Either<Self, Self> ⓘ
self into a Left variant of Either<Self, Self>
if into_left is true.
Converts self into a Right variant of Either<Self, Self>
otherwise. Read moreSource§fn into_either_with<F>(self, into_left: F) -> Either<Self, Self> ⓘ
fn into_either_with<F>(self, into_left: F) -> Either<Self, Self> ⓘ
self into a Left variant of Either<Self, Self>
if into_left(&self) returns true.
Converts self into a Right variant of Either<Self, Self>
otherwise. Read more