A high-resolution spectral ultrafast compressed coherent modulation imaging system and method

Abstract

The application discloses a kind of high-resolution spectral ultrafast compressed coherent modulation imaging system and method, wherein high-resolution spectral ultrafast compressed coherent modulation imaging system includes light-receiving module and lensless encoding imaging module sequentially arranged along the direction of propagation of light path, and lensless encoding imaging module includes transmission grating, phase encoding board and stripe camera.Scattered light is imaged to rear end image plane by light-receiving module, then sequentially generates spectral space shear in free space by transmission grating, carries out spatial phase modulation by phase encoding board, and is realized time space scanning record by stripe camera with slit completely open, obtains two-dimensional intensity image.High-resolution spectral ultrafast compressed coherent modulation imaging method is based on the forward observation model including free space diffraction, spectral shear, phase modulation and time scanning, reconstructs the two-dimensional space, time, wavelength and phase information of dynamic complex amplitude light field to be measured from two-dimensional intensity image.

Description

Technical Field

[0001] This invention relates to the field of laser inertial confinement fusion technology, specifically to a high-resolution spectral ultrafast compressed coherent modulation imaging system and method. Background Technology

[0002] Due to the enclosed nature of black cavities and the complexity of the environment in laser inertial confinement fusion (ICF) experiments, the main methods for diagnosing the state of black cavity plasmas are optical Thomson scattering (OTS) and X-ray spectroscopy. Optical Thomson scattering is an active diagnostic method that uses a probe light of a specific wavelength to drive the oscillation of directed electrons, thereby generating scattered light with a Doppler frequency shift effect. By analyzing the spectrum of the scattered light, the state evolution processes of the plasma, such as electron temperature, electron density, plasma velocity, and ion temperature, can be inferred.

[0003] Existing optical Thomson scattering diagnostic techniques mainly include imaging-based optical Thomson scattering diagnostics and stripe-based optical Thomson scattering diagnostics. Imaging-based optical Thomson scattering diagnostics typically projects the scattered light through an imaging mirror to a spectrometer slit. After dispersion by the spectrometer, the image is recorded by an array detector, thus obtaining the spectral information of the scattered light at a one-dimensional spatial location within the plasma. Stripe-based optical Thomson scattering diagnostics typically employs a diagnostic system combining a spectrometer and a stripe camera. The imaged scattered light is fed into a spectrometer-stripe camera system arranged vertically in the slit, thereby recording the local spectral characteristics of the scattered light changing over time within a specific spatial region, enabling time-resolved measurements of electron density and temperature.

[0004] However, imaging-based optical Thomson scattering diagnostics typically only obtains spectral information corresponding to a one-dimensional spatial location, making it difficult to simultaneously acquire complete two-dimensional spatial distribution and temporal evolution information. While fringe-based optical Thomson scattering diagnostics can obtain time-resolved spectral information, it is usually limited to a local spatial region, resulting in limited spatial information dimensions. All of these traditional methods struggle to simultaneously acquire two-dimensional spatial, temporal, spectral, and phase evolution information of scattered light in a single or limited number of acquisitions.

[0005] For the rapidly evolving, strongly coupled, and multi-scale plasma processes in inertial confinement fusion experiments, relying solely on low-dimensional spectral measurements is insufficient to meet the demands of refined physical modeling, simulation correction, and experimental optimization. Therefore, there is an urgent need for an ultrafast compressed imaging system and reconstruction method capable of simultaneously acquiring spatiotemporally resolved complex amplitude information of scattered light in a single or limited number of acquisitions. Summary of the Invention

[0006] To address the technical problems of insufficient information acquisition dimensions and difficulty in simultaneously recording temporal, spatial, spectral, and phase evolution information in existing optical Thomson scattering diagnostic techniques, this invention provides a high-resolution spectral ultrafast compressed coherent modulation imaging system and method, which realizes the simultaneous acquisition and reconstruction of five-dimensional information of the evolution process of transient incoherent light sources, including time, two-dimensional space, wavelength, and phase.

[0007] The technical solution is as follows:

[0008] The first aspect of this application relates to a high-resolution spectral ultrafast compressed coherent modulation imaging system, comprising a light-collecting module and a lensless coded imaging module arranged sequentially along the optical path propagation direction.

[0009] The light-collecting module is used to collect scattered light carrying information about ultrafast dynamic processes and image the scattered light onto the back-end image plane.

[0010] The lensless coded imaging module includes a transmissive grating, a phase coding plate, and a stripe camera arranged sequentially along the light propagation direction.

[0011] The rear image plane is located between the light-receiving module and the transmissive grating, and there is a first propagation distance between the rear image plane and the transmissive grating. The transmissive grating and the phase encoding plate have a second propagation distance. The phase encoder and the stripe camera have a third propagation distance. ;

[0012] The dynamic complex amplitude light field to be measured at the rear image plane propagates through free-space paraxial near-field diffraction over a first propagation distance. Then it reaches the transmission grating, which is used to generate wavelength-related lateral dispersion shifts in a plane perpendicular to the optical axis for different wavelength components.

[0013] The light field modulated by the aforementioned transmissive grating propagates through paraxial near-field diffraction in free space for a second propagation distance. Then it reaches the phase encoding board, which is used to perform spatial phase encoding modulation of the light field in a plane perpendicular to the optical axis;

[0014] The light field modulated by the phase encoding plate propagates through paraxial near-field diffraction in free space for a third propagation distance. Then it reaches the streak camera, whose cathode slit is fully open, used to scan and unfold the incident light field in a plane perpendicular to the optical axis according to the time dimension, and record a two-dimensional intensity image containing the multiplexing results of two-dimensional spatial, temporal, spectral and phase information.

[0015] The above-described high-resolution spectral ultrafast compressed coherent modulation imaging system maps the spectral dimension to a spatial lateral displacement using a transmission grating and the temporal dimension to a spatial longitudinal displacement using a streak camera. Furthermore, it introduces computable and decodeable spatial phase modulation through a phase encoding plate and free-space diffraction propagation, thus compressing and encoding the high-dimensional information of the measured dynamic complex amplitude light field into a two-dimensional intensity image. Compared to traditional imaging-type and streak-type optical Thomson scattering diagnostics, this invention can obtain higher-dimensional information in one or fewer acquisitions, providing a physical data foundation for spatiotemporally spectrally resolved complex amplitude reconstruction of scattered light.

[0016] In some embodiments, the light-collecting module includes a first lens group and a second lens group; the first lens group is used to collect incoherent scattered light and collimate it into a parallel beam; the second lens group is used to converge the parallel beam and focus it onto the rear image plane.

[0017] With the above design, the first lens group can improve the collection efficiency of weak scattered light, which is especially suitable for Thomson scattered light with large divergence angle and weak light intensity in inertial confinement fusion experiments; the second lens group can adjust the imaging magnification and numerical aperture to form a dynamic complex amplitude light field with high spatial resolution and signal-to-noise ratio at the rear image plane, thereby improving the coding quality and reconstruction accuracy of the subsequent lensless coding imaging module.

[0018] In some embodiments, the dispersion direction of the transmissive grating intersects or is orthogonal to the temporal scanning direction of the streak camera in a plane perpendicular to the optical axis, such that the spectral and temporal dimensions of the light field are cut and encoded along different spatial directions.

[0019] By adopting the above design, the spectral-spatial shearing direction of the transmission grating intersects or is orthogonal to the temporal-spatial shearing direction of the streak camera. This can reduce the degree of aliasing of the spectral and temporal dimensions on the recording surface, allowing spectral and temporal information to unfold along different spatial directions. This is beneficial for improving the resolvability and reconstruction stability of compressed observation data.

[0020] In some embodiments, the phase encoding plate is a quartz substrate, on which etched pixel blocks are randomly distributed. The etched pixel blocks have a preset etching depth and are used to apply a spatially non-uniform phase delay to the incident light field.

[0021] Using the above design, randomly distributed etched pixel blocks can apply a spatially non-uniform phase delay to the incident light field, increasing the randomness and diversity of the system's measurement matrix. This allows information from different spatial locations, wavelengths, and times to be fully mixed during diffraction propagation, thereby improving the solvability of the compression reconstruction problem. The quartz substrate has high transmittance, good thermal stability, and strong resistance to laser damage, making it suitable for high-power laser diagnostic environments.

[0022] The second aspect of this application relates to a high-resolution spectral ultrafast compressed coherent modulation imaging method, characterized in that, based on the above-mentioned high-resolution spectral ultrafast compressed coherent modulation imaging system, it includes the following steps:

[0023] S1. Construct a forward observation model to describe the process by which the dynamic complex amplitude light field to be measured at the back-end image plane forms a two-dimensional intensity image after passing through a transmission grating, a phase encoding plate, free space diffraction propagation, and spatiotemporal scanning by a streak camera.

[0024] S2. Acquire two-dimensional intensity images using a streak camera with the cathode slit fully open.

[0025] S3. Based on the forward observation model, the two-dimensional intensity image is compressed and reconstructed to obtain the amplitude and phase information of the dynamic complex amplitude light field to be measured in two-dimensional space, time and wavelength dimensions.

[0026] By employing the high-resolution spectral ultrafast compressed coherent modulation imaging method described above, and establishing a forward observation model corresponding to the actual optical system, combined with compressed reconstruction of two-dimensional intensity images, a high-dimensional dynamic complex amplitude light field can be recovered from low-dimensional detection data. This method solves the problem that streak cameras can only directly record two-dimensional intensity images and cannot directly record phase, while also alleviating the problem of insufficient information dimension in traditional diagnostic methods.

[0027] In some implementations, in step S1, the forward observation model is represented as:

[0028] ;

[0029] In the above formula, Represents a two-dimensional intensity image. Describes the spatiotemporal integral operator of the streak camera. This represents the spatiotemporal shearing operator of the stripe camera. This represents the spectral spatial shearing operator on the recording plane of the streak camera. This represents the phase modulation operator of the phase encoder board. Indicates the first propagation distance The spatial free propagation operator, Indicates the second propagation distance The spatial free propagation operator, Indicates the third propagation distance The spatial free propagation operator, Represents a dynamic complex amplitude optical field, where, and The x and y coordinates represent the two-dimensional spatial information of the dynamic complex amplitude light field. This represents the temporal information of the dynamic complex amplitude optical field. This represents the wavelength information of the dynamic complex amplitude optical field. It represents the phase information of the dynamic complex amplitude optical field.

[0030] Using the above method, the forward observation model unifies the representation of the transmission grating, phase encoding plate, free space propagation, and the striated camera scanning process, enabling the reconstruction algorithm to accurately utilize the physical encoding laws of the system, thereby improving the accuracy of complex amplitude reconstruction. The model explicitly includes an absolute value squaring operation, which accurately reflects the physical fact of the striated camera recording intensity rather than the complex amplitude.

[0031] In some implementations, step S3 employs the GS-GAP algorithm for compression and reconstruction, the GS-GAP algorithm comprising the following steps:

[0032] Define the forward complex amplitude field propagation operator Define the observation operator Solving the forward observation model can be formulated as a constrained optimization problem with total variational priors as regularization constraints:

[0033] ;

[0034] In the above formula, Represents the dynamic complex amplitude light field , Indicates to Total variational regularization, Represents a two-dimensional intensity image , Indicates to make Find the minimum value and satisfy the constraints At that time, the corresponding The value of ;

[0035] Introducing auxiliary variables into the generalized alternating projection framework The constrained optimization problem is decoupled into a data fidelity subproblem and a prior constraint subproblem.

[0036] In the In the iteration, based on the first Auxiliary variable estimates in the next iteration Perform the data fidelity step: As a unitary operator that guarantees norm invariance, let Let the current estimated forward projection onto the target surface be... The data fidelity subproblem is transformed into:

[0037] ;

[0038] In the above formula, Indicates to make Find the minimum value and satisfy the constraints At that time, the corresponding The value of ;

[0039] According to the Gerchberg-Saxton algorithm principle, while satisfying the target strength constraint, the phase of the current iteration is kept unchanged and the amplitude is replaced, thus reducing the Euclidean distance minimization in the complex domain to a pure amplitude minimization problem.

[0040] Introducing the previous light intensity distribution estimate and the updated light intensity distribution estimate For the problem of minimizing pure amplitude Perform a first-order Taylor expansion and define the diagonal weight matrix. The nonlinear phase retrieval problem is transformed into an estimate of the updated light intensity distribution. The weighted least squares problem;

[0041] A Lagrangian function is constructed for the weighted least squares problem, and the updated light intensity distribution estimate is obtained by solving it. Exact closed-form solution:

[0042] ;

[0043] In the above formula, This represents the transpose of the measurement matrix of the observation operator;

[0044] The exact closed-form solution is assigned back to the phase predicted during the forward propagation, and then backpropagated to obtain the first... The dynamic complex amplitude optical field estimate of the next iteration Update formula:

[0045] ;

[0046] In the above formula, The inverse operator represents the forward complex amplitude field propagation operator;

[0047] Execute the prior constraint step: for Perform total variation denoising and update the estimated values ​​of auxiliary variables. ;

[0048] The data fidelity step and the prior constraint step are executed alternately until the residual convergence threshold is met or the maximum number of iterations is reached, and the final converged dynamic complex amplitude optical field estimate is output.

[0049] Using the above methods, the GS-GAP algorithm combines the nonlinear phase retrieval problem with the compressed sensing reconstruction problem. On the one hand, it maintains the current phase and updates the amplitude under intensity constraints through the Gerchberg-Saxton idea. On the other hand, it introduces prior constraints through the GAP framework, so that the underdetermined nonlinear inverse problem can be decomposed into data fidelity subproblems and prior constraint subproblems that are solved alternately, thereby improving the stability and accuracy of reconstruction. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of a high-resolution spectral ultrafast compressed coherent modulation imaging system.

[0051] Figure 2 A schematic diagram of the optical path for applying a high-resolution spectral ultrafast compressed coherent modulation imaging system to collective Thomson scattering diagnosis in inertial confinement fusion.

[0052] Figure 3 A schematic diagram showing the amplitude reconstruction results using the GS-GAP algorithm;

[0053] Figure 4 This is a schematic diagram of the phase reconstruction results using the GS-GAP algorithm.

[0054] Explanation of reference numerals in the attached figures:

[0055] 1. Light receiving module; 2. Lensless coded imaging module; 3. Transmissive grating; 4. Phase coding plate; 5. Stripe camera; 6. Rear image plane; 7. First lens group; 8. Second lens group; 9. Target chamber; 10. Target surface. Detailed Implementation

[0056] The present invention will be further described below with reference to the embodiments and accompanying drawings.

[0057] Example 1:

[0058] like Figure 1 As shown, this embodiment provides a high-resolution spectral ultrafast compressed coherent modulation imaging system, which mainly includes a light-receiving module 1 and a lensless coded imaging module 2 arranged sequentially along the optical path propagation direction.

[0059] The light receiving module 1 is used to collect scattered light carrying information about ultrafast dynamic processes and image the scattered light onto the rear image plane 6, where the dynamic complex amplitude light field to be measured is formed.

[0060] The lensless coded imaging module 2 includes a transmissive grating 3, a phase coding plate 4, and a stripe camera 5 arranged sequentially along the light propagation direction.

[0061] The rear image plane 6 is located between the light-receiving module 1 and the transmissive grating 3. The distance between the rear image plane 6 and the transmissive grating 3 is the first propagation distance. The distance between the transmissive grating 3 and the phase encoding plate 4 is the second propagation distance. The distance between phase encoder 4 and stripe camera 5 is the third propagation distance. .

[0062] In this embodiment, the lensless coded imaging module 2 is the core component for achieving multidimensional information compression recording. Lensless coded imaging refers to the process of directly imaging the rear image plane 6 onto the detector surface without relying on traditional imaging lens groups between the rear image plane 6 and the streak camera 5. Instead, it utilizes free-space diffraction propagation, spectral dispersion, spatial phase modulation, and temporal scanning to jointly encode and record the high-dimensional dynamic complex amplitude light field.

[0063] 1. Propagation from the rear image plane to the transmission grating

[0064] The dynamic complex amplitude light field emitted from the rear image plane 6 first propagates through paraxial near-field diffraction in free space over a first propagation distance. Then it reaches the transmission grating 3.

[0065] 2. Spectral-spatial shearing of transmission gratings

[0066] When the light field reaches the transmission grating 3, different wavelength components produce different diffraction angles, thereby generating a wavelength-dependent lateral dispersion shift in a plane perpendicular to the optical axis. This process can be understood as mapping the spectral dimension to the lateral direction of space, forming a spectral-spatial shearing.

[0067] 3. Propagation from the transmissive grating to the phase encoder board

[0068] The light field modulated by the transmission grating 3 continues to propagate paraxially in free space via near-field diffraction at a second propagation distance. It reaches phase encoder 4. Second propagation distance The configuration allows for further diffraction and expansion of the different wavelength components after dispersion by the transmission grating 3, forming a spatially and spectrally related complex amplitude distribution at the phase encoding plate 4. A reasonable selection of the second propagation distance is crucial. It can enhance the coupling between spectral shearing and spatial coding, and improve the diversity of compressed measurements.

[0069] 4. Spatial phase modulation of the phase encoder board

[0070] The phase encoding board 4 is used to perform spatial phase encoding modulation of the incident light field in a plane perpendicular to the optical axis. In this embodiment, the phase encoding board 4 is a quartz substrate with randomly distributed etched pixel blocks, each with a preset etching depth. The etching depth can be designed according to the center wavelength to produce a phase delay at the center wavelength, or it can be designed to produce other preset phase delays. The phase encoding board 4 can be a binary phase encoding board or a multi-level phase encoding board. The randomly distributed pixel blocks cause different spatial positions of the incident light field to experience different phase delays, thereby forming a complex but calibrable diffraction mixture in subsequent propagation. Since the phase encoding function is known, it can be modeled and inverted in the reconstruction algorithm using the phase modulation operator of the phase encoding board 4.

[0071] 5. Propagation from the phase encoder to the stripe camera

[0072] The light field modulated by phase encoder 4 continues to propagate paraxially near-field diffraction in free space for the third propagation distance. It reaches the cathode surface of the streak camera 5. Third propagation distance. This is used to further diffract and mix the phase-modulated light field, converting the phase perturbation introduced by the phase encoding plate 4 into diffraction intensity changes that can be recorded by the intensity image. This is achieved by appropriately setting the third propagation distance. This can improve the sensitivity of the conversion of phase information into intensity information, thus aiding in subsequent phase reconstruction.

[0073] 6. Temporal-spatial shearing and intensity recording by a stripe camera

[0074] The cathode slit of the streak camera 5 is in a fully open state. Unlike traditional streak cameras that record one-dimensional space or local area information only through the slit, in this embodiment, the streak camera 5 receives the two-dimensional incident light field with the slit fully open, and uses the internal scanning deflection electric field to deflect photons arriving at different times to different positions in the vertical direction, thereby achieving time-space shearing.

[0075] Finally, the streak camera 5 records a two-dimensional intensity image. This two-dimensional intensity image simultaneously reuses the two-dimensional spatial, temporal, spectral, and phase information of the light field under test. Since the detector can only record intensity, the recording process includes a complex amplitude mode squaring operation.

[0076] Through the aforementioned optical path, the transmissive grating 3 is responsible for shearing the spectral dimension to the horizontal direction in space, the streak camera 5 is responsible for shearing the temporal dimension to the vertical direction in space, and the phase encoding plate 4 and free-space diffraction propagation together achieve the encoding and mixing of spatial structure and phase information. Therefore, this system can provide sufficient encoding information for multidimensional complex amplitude reconstruction from two-dimensional intensity detection data.

[0077] In this embodiment, the light receiving module 1 includes a first lens group 7 and a second lens group 8.

[0078] The first lens group 7 is used to collect scattered light and collimate it into a parallel beam. For Thomson scattered light in inertial confinement fusion experiments, since the scattered light is usually weak and has a certain divergence angle, the first lens group 7 preferably uses a large-aperture lens combination to improve light collection efficiency. The focal length, aperture, and numerical aperture of the first lens group 7 can be designed according to the divergence angle of the scattered light, the size of the target area, and the experimental space constraints.

[0079] The second lens group 8 is used to converge the parallel beam output from the first lens group 7 and image it onto the rear image plane 6. By adjusting the focal length and position of the second lens group 8, the imaging magnification, field of view, and spatial sampling interval on the rear image plane 6 can be controlled. In one embodiment, the first lens group 7 and the second lens group 8 can together form a telescopic imaging structure, so that the scattered light forms a real image of the target area at the rear image plane 6 after passing through the light receiving module 1. The light field at the rear image plane 6 contains both spatial distribution information and retains temporal, wavelength, and phase evolution information.

[0080] The phase encoding board 4 is preferably a random phase encoding board. Specifically, the phase encoding board 4 can be a quartz substrate, on which randomly distributed etched pixel blocks are prepared using processes such as photolithography, wet etching, dry etching, or laser writing. Each etched pixel block can have the same depth, or it can be designed to have different depths as needed. The shape of the pixel block can be square, circular, polygonal, or other regular or irregular shapes.

[0081] The pixel size of the phase encoder 4 is preferably matched with the system spatial resolution and the sampling pixels of the stripe camera. For example, when the rear image plane 6 propagates to the phase encoder 4 through free space and the light field spatial sampling interval is a set value, the pixel size of the phase encoder 4 can be set to one or several times the above-mentioned set value to balance modulation intensity and processing feasibility.

[0082] The random distribution of the phase encoder 4 can be a binary random distribution, for example, each pixel can be an etched area or a non-etched area with a preset probability; or it can be a pseudo-random coding distribution, such as generated based on a Bernoulli random sequence, an M-sequence, or other preset random patterns. As long as the coding distribution of the phase encoder 4 is known in the reconstruction algorithm, it can be used to construct the phase modulation operator.

[0083] Please see Figure 2 When the high-resolution spectral ultrafast compressed coherent modulation imaging system of this embodiment is applied to the diagnosis of collective Thomson scattering in inertial confinement fusion, the first lens group 7 is set at the outlet of the target chamber 9. After the probe light is focused on the target surface 10, the Thomson scattered light is led out of the target chamber 9 through the first lens group 7.

[0084] For practical experimental scenarios, several reflectors can be placed between the first lens group 7 and the second lens group 8.

[0085] Example 2:

[0086] A high-resolution spectral ultrafast compressed coherent modulation imaging system, comprising the following steps:

[0087] S1. Construct a forward observation model to describe the process by which the dynamic complex amplitude light field to be measured at the back-end image plane is transformed into a two-dimensional intensity image after passing through a transmission grating 3, a phase encoding plate 4, free space diffraction propagation, and a streak camera 5 for spatiotemporal scanning.

[0088] Specifically, the dynamic complex amplitude light field emitted from the rear image plane 6 It involves two-dimensional spatial information, time, wavelength, and phase information, among which, and The x and y coordinates represent the two-dimensional spatial information of the dynamic complex amplitude light field. This represents the temporal information of the dynamic complex amplitude optical field. This represents the wavelength information of the dynamic complex amplitude optical field. It represents the phase information of the dynamic complex amplitude optical field.

[0089] First, the dynamic complex amplitude light field After the first transmission distance The propagation reaches the transmission grating 3, and the spectral dimension is deflected in the horizontal direction of space.

[0090] Then, after the second propagation distance The propagation of light is modulated by phase encoding plate 4.

[0091] Finally, after propagating the third propagation distance Subsequently, under the action of the streak camera 5 with the cathode slit fully open, the light field underwent a temporal deflection in the vertical direction of space and was ultimately recorded as a two-dimensional intensity image. .

[0092] Therefore, the forward observation model is expressed as:

[0093] ;

[0094] In the above formula, Describes the spatiotemporal integral operator of the streak camera. This represents the spatiotemporal shearing operator of the stripe camera. This represents the spectral spatial shearing operator on the recording plane of the streak camera. This represents the phase modulation operator of the phase encoder board. Indicates the first propagation distance The spatial free propagation operator, Indicates the second propagation distance The spatial free propagation operator, Indicates the third propagation distance The space-free propagation operator.

[0095] S2. Use the stripe camera 5 to acquire a two-dimensional intensity image with the cathode slit fully open.

[0096] Furthermore, in order to improve the reconstruction accuracy in actual experiments, the system parameters are calibrated before performing step S2, including the following steps:

[0097] 1. Spectral-spatial shear calibration

[0098] Multiple narrowband light sources or tunable lasers with known center wavelengths are sequentially incident on the system. The lateral positions of different wavelengths on the recording surface of the streak camera are recorded, and the correspondence between wavelength and lateral displacement is obtained. From this, the spectral spatial shearing operator on the recording surface of the streak camera is determined. .

[0099] 2. Time-space shear calibration

[0100] An ultrashort pulse sequence with known time intervals is incident on a streak camera. The longitudinal position of the pulses on the recording surface at different times is recorded, and the correspondence between time and longitudinal displacement is obtained. Based on this, the spatiotemporal shearing operator of the streak camera is determined. .

[0101] 3. Phase encoder board calibration

[0102] The coding distribution of the phase coding board can be obtained directly from the design documents, or it can be calibrated using interferometry, profilometry, or phase retrieval methods. After calibration, the phase coding function is obtained, and a phase modulation operator is constructed based on it.

[0103] 4. Propagation distance and system sampling calibration

[0104] The first propagation distance is determined by mechanical measurement or optical calibration. Second transmission distance Third transmission distance Simultaneously, the computational grid size is determined based on the stripe camera pixel size, system magnification, and light field sampling requirements. This is used to construct the first propagation distance. Free propagation operator in space Second transmission distance Free propagation operator in space Third transmission distance Free propagation operator in space .

[0105] The above calibration process can reduce the impact of systematic errors on the complex amplitude reconstruction results and improve the reliability of actual diagnostic data.

[0106] S3. Based on the forward observation model, the two-dimensional intensity image is compressed and reconstructed to obtain the amplitude and phase information of the dynamic complex amplitude light field to be measured in two-dimensional space, time and wavelength dimensions.

[0107] Specifically, the GS-GAP algorithm is used for compression and reconstruction, and the GS-GAP algorithm includes the following steps:

[0108] From the perspective of solving the mathematical model, the dimension of unknowns is much larger than the dimension of measured values, and it involves the operation of squaring absolute values. Therefore, this is a typical nonlinear problem of finding the inverse of an underdetermined system of equations. Direct solution is an extremely ill-conditioned inverse problem, and prior knowledge must be introduced for regularization constraints. For simplicity, we define a forward complex amplitude field propagation operator. Define the observation operator Simultaneously, a total variation (TV) prior is introduced, and the solution to the forward observation model is formulated as a constrained optimization problem with a TV prior as a regularization constraint:

[0109] ;

[0110] In the above formula, Represents the dynamic complex amplitude light field , Indicates to Total variational regularization, Represents a two-dimensional intensity image , Indicates to make Find the minimum value and satisfy the constraints At that time, the corresponding The value of ;

[0111] Introducing auxiliary variables into the generalized alternating projection (GAP) framework The constrained optimization problem is decoupled into a data fidelity subproblem and a prior constraint subproblem.

[0112] In the In the iteration, based on the first Auxiliary variable estimates in the next iteration Perform the data fidelity step: As a unitary operator that guarantees norm invariance, the optimization process can therefore be transformed from the objective domain to the target domain. Let Let the current estimated forward projection onto the target surface be... The data fidelity subproblem is transformed into:

[0113] ;

[0114] In the above formula, Indicates to make Find the minimum value and satisfy the constraints At that time, the corresponding The value of ;

[0115] The problem described above is a highly non-convex problem, which needs to be transformed into a convex optimization problem that can be solved exactly.

[0116] Based on the principles of the Gerchberg-Saxton (GS) algorithm, while satisfying the target strength constraint, the phase of the current iteration is kept unchanged during amplitude replacement, thus reducing the Euclidean distance minimization problem in the complex domain to a pure amplitude minimization problem. Let:

[0117] ;

[0118] In the above formula, It represents the Hadamah accumulation. Represents the imaginary unit;

[0119] At this point, the minimization of the Euclidean distance of the complex number degenerates into the minimization of the pure amplitude:

[0120] ;

[0121] Introducing the previous light intensity distribution estimate and the updated light intensity distribution estimate That is: the updated light intensity distribution estimate It is already very close to the true value. This applies to the problem of minimizing the pure amplitude. Perform a first-order Taylor expansion:

[0122] ;

[0123] Substituting this expansion into the objective function, the nonlinear square root term is simplified to an estimate of the light intensity distribution before the update. Linear polynomial:

[0124] ;

[0125] Ignore the constant coefficient 1 / 4, which does not affect the location of the extreme value, and define a diagonal weight matrix. The nonlinear phase retrieval problem is transformed into an estimate of the updated light intensity distribution. Weighted least squares problem:

[0126] ;

[0127] For the above quadratic programming problem, construct the Lagrange function:

[0128] ;

[0129] The updated light intensity distribution estimate is obtained by solving the problem. Exact closed-form solution:

[0130] ;

[0131] In the above formula, This represents the transpose of the measurement matrix of the observation operator;

[0132] The exact closed-form solution is assigned back to the phase predicted during the forward propagation, and then backpropagated to obtain the first... The dynamic complex amplitude optical field estimate of the next iteration Update formula:

[0133] ;

[0134] In the above formula, The inverse operator represents the forward complex amplitude field propagation operator;

[0135] Execute the prior constraint step: for Perform total variation denoising and update the estimated values ​​of auxiliary variables. ;

[0136] The data fidelity step and the prior constraint step are executed alternately until the residual convergence threshold is met or the maximum number of iterations is reached, and the final converged dynamic complex amplitude optical field estimate is output.

[0137] Please see Figure 3 and Figure 4 To verify the feasibility of the method of this invention, numerical simulations can be performed on a high-resolution spectral ultrafast compressed coherent modulation imaging system. In one simulation implementation, different water droplet patterns are used as the amplitude and phase distributions of the dynamic complex amplitude light field to simulate the spatiotemporal spectral evolution of the transient light field. The dynamic process is divided into four time points, t1 to t4, and the simulated pattern is rotated sequentially at each time point to represent the change of the dynamic scene over time. The spectral range is uniformly divided into... to Five wavelength slices are used, with the light field intensity of each slice arranged according to a normal distribution to simulate the intensity distribution of the actual Thomson scattering spectrum.

[0138] During the simulation, firstly according to the set... , , A forward observation model is constructed using the phase encoder distribution, grating dispersion coefficient, and fringe camera scanning coefficient. A two-dimensional compressed intensity image is then generated based on this model. Subsequently, the two-dimensional intensity image is used as input to reconstruct the complex amplitude using the GS-GAP algorithm.

[0139] The reconstruction results can be displayed as amplitude maps and phase maps, respectively. For each time point and each wavelength slice, the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) between the reconstructed result and the original simulated pattern are calculated.

[0140] In one simulation result, the PSNR value of the reconstructed amplitude map ranged from 32.0 dB to 33.9 dB, with an average of approximately 33.1 dB, and the SSIM value ranged from 0.980 to 0.985, with an average of approximately 0.983. The PSNR value of the reconstructed phase map ranged from 32.8 dB to 35.1 dB, with an average of approximately 33.9 dB, and the SSIM value ranged from 0.967 to 0.978, with an average of approximately 0.972. These results demonstrate that the system and GS-GAP algorithm proposed in this invention can accurately recover the amplitude and phase information of the dynamic complex amplitude optical field from two-dimensional intensity compression measurement data.

[0141] Therefore, the high-resolution spectral ultrafast compressed coherent modulation imaging system and method proposed in this invention acquires the complex amplitude light field of the back-end image plane of the scattered light under test through a light-receiving module, realizes spectral-spatial shearing, spatial phase encoding, free-space diffraction mixing, and temporal-spatial scanning recording through a lensless coded imaging module, and recovers the dynamic complex amplitude information from the two-dimensional intensity image through the GS-GAP algorithm. This scheme can acquire two-dimensional spatial, temporal, and spectral resolved amplitude and phase information of scattered light in a single or few acquisitions, and is suitable for collective Thomson scattering diagnosis in inertial confinement fusion experiments, and can also be extended to other ultrafast transient light field imaging scenarios.

[0142] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention. Those skilled in the art, under the guidance of the present invention, can make various similar representations without departing from the spirit and claims of the present invention, and such modifications all fall within the protection scope of the present invention.

Claims

1. A high-resolution spectral ultrafast compressed coherent modulation imaging system, characterized in that, It includes a light-collecting module and a lensless coded imaging module arranged sequentially along the optical path propagation direction; The light-collecting module is used to collect scattered light carrying information about ultrafast dynamic processes and image the scattered light onto the back-end image plane. The lensless coded imaging module includes a transmissive grating, a phase coding plate, and a stripe camera arranged sequentially along the light propagation direction. The rear image plane is located between the light-receiving module and the transmissive grating, and there is a first propagation distance between the rear image plane and the transmissive grating. The transmissive grating and the phase encoding plate have a second propagation distance. The phase encoder and the stripe camera have a third propagation distance. ; The dynamic complex amplitude light field to be measured at the rear image plane propagates through free-space paraxial near-field diffraction over a first propagation distance. Then it reaches the transmission grating, which is used to generate wavelength-related lateral dispersion shifts in a plane perpendicular to the optical axis for different wavelength components. The light field modulated by the aforementioned transmissive grating propagates through paraxial near-field diffraction in free space for a second propagation distance. Then it reaches the phase encoding board, which is used to perform spatial phase encoding modulation of the light field in a plane perpendicular to the optical axis; The light field modulated by the phase encoding plate propagates through paraxial near-field diffraction in free space for a third propagation distance. Then it reaches the streak camera, whose cathode slit is fully open, used to scan and unfold the incident light field in a plane perpendicular to the optical axis according to the time dimension, and record a two-dimensional intensity image containing the multiplexing results of two-dimensional spatial, temporal, spectral and phase information.

2. The high-resolution spectral ultrafast compressed coherent modulation imaging system according to claim 1, characterized in that, The light-collecting module includes a first lens group and a second lens group; the first lens group is used to collect incoherent scattered light and collimate it into a parallel beam; the second lens group is used to converge the parallel beam and focus it onto the rear image plane.

3. The high-resolution spectral ultrafast compressed coherent modulation imaging system according to claim 1, characterized in that, The dispersion direction of the transmissive grating intersects or is orthogonal to the temporal scanning direction of the streak camera in a plane perpendicular to the optical axis, so that the spectral and temporal dimensions of the light field are cut and encoded along different spatial directions.

4. The high-resolution spectral ultrafast compressed coherent modulation imaging system according to claim 1, characterized in that, The phase encoding board is a quartz substrate, on which etched pixel blocks are randomly distributed. The etched pixel blocks have a preset etching depth and are used to apply a spatially non-uniform phase delay to the incident light field.

5. A high-resolution spectral ultrafast compressed coherent modulation imaging method, characterized in that, The high-resolution spectral ultrafast compressed coherent modulation imaging system based on any one of claims 1 to 4 includes the following steps: S1. Construct a forward observation model to describe the process by which the dynamic complex amplitude light field to be measured at the back-end image plane forms a two-dimensional intensity image after passing through a transmission grating, a phase encoding plate, free space diffraction propagation, and spatiotemporal scanning by a streak camera. S2. Acquire two-dimensional intensity images using a streak camera with the cathode slit fully open. S3. Based on the forward observation model, the two-dimensional intensity image is compressed and reconstructed to obtain the amplitude and phase information of the dynamic complex amplitude light field to be measured in two-dimensional space, time and wavelength dimensions.

6. The high-resolution spectral ultrafast compressed coherent modulation imaging method according to claim 5, characterized in that, In step S1, the forward observation model is represented as follows: ； In the above formula, Represents a two-dimensional intensity image. Describes the spatiotemporal integral operator of the streak camera. This represents the spatiotemporal shearing operator of the stripe camera. This represents the spectral spatial shearing operator on the recording plane of the streak camera. This represents the phase modulation operator of the phase encoder board. Indicates the first propagation distance The spatial free propagation operator, Indicates the second propagation distance The spatial free propagation operator, Indicates the third propagation distance The spatial free propagation operator, Represents a dynamic complex amplitude optical field, where, and The x and y coordinates represent the two-dimensional spatial information of the dynamic complex amplitude light field. This represents the temporal information of the dynamic complex amplitude optical field. This represents the wavelength information of the dynamic complex amplitude optical field. It represents the phase information of the dynamic complex amplitude optical field.

7. The high-resolution spectral ultrafast compressed coherent modulation imaging method according to claim 6, characterized in that, Step S3 uses the GS-GAP algorithm for compression and reconstruction. The GS-GAP algorithm includes the following steps: Define the forward complex amplitude field propagation operator Define the observation operator Solving the forward observation model can be formulated as a constrained optimization problem with total variational priors as regularization constraints: ； In the above formula, Represents the dynamic complex amplitude light field , Indicates to Total variational regularization, Represents a two-dimensional intensity image , Indicates to make Find the minimum value and satisfy the constraints At that time, the corresponding The possible values ​​of ; Introducing auxiliary variables into the generalized alternating projection framework The constrained optimization problem is decoupled into a data fidelity subproblem and a prior constraint subproblem. In the In the iteration, based on the first Auxiliary variable estimates in the next iteration Perform the data fidelity step: As a unitary operator that guarantees norm invariance, let Let the current estimated forward projection onto the target surface be... The data fidelity subproblem is transformed into: ； In the above formula, Indicates to make Find the minimum value and satisfy the constraints At that time, the corresponding The possible values ​​of ; According to the Gerchberg-Saxton algorithm principle, while satisfying the target strength constraint, the phase of the current iteration is kept unchanged and the amplitude is replaced, thus reducing the Euclidean distance minimization in the complex domain to a pure amplitude minimization problem. Introducing the previous light intensity distribution estimate and the updated light intensity distribution estimate For the problem of minimizing pure amplitude Perform a first-order Taylor expansion and define the diagonal weight matrix. The nonlinear phase retrieval problem is transformed into an estimate of the updated light intensity distribution. The weighted least squares problem; A Lagrangian function is constructed for the weighted least squares problem, and the updated light intensity distribution estimate is obtained by solving it. Exact closed-form solution: ； In the above formula, This represents the transpose of the measurement matrix of the observation operator; The exact closed-form solution is assigned back to the phase predicted during the forward propagation, and then backpropagated to obtain the first... The dynamic complex amplitude light field estimate of the next iteration Update formula: ； In the above formula, The inverse operator represents the forward complex amplitude field propagation operator; Execute the prior constraint step: for Perform total variation denoising and update the estimated values ​​of auxiliary variables. ; The data fidelity step and the prior constraint step are executed alternately until the residual convergence threshold is met or the maximum number of iterations is reached, and the final converged dynamic complex amplitude optical field estimate is output.

Images