Coherent snapshot compression imaging method, storage medium and device for high-dimensional signals

Abstract

The present invention discloses a coherent snapshot compression imaging method, storage medium, and device for high-dimensional signals. The method comprises the following steps: S1: modulating the space-light field propagation in a coherent light illumination scene using a composite phase mask generated by a spatial light modulator; S2: superimposing and recording the images on a detector plane to produce a single hologram generated by the space-light field compression modulation obtained from a single exposure; and S3: transmitting the hologram to a receiver and reconstructing the original signal using an existing reconstruction algorithm to achieve coherent snapshot compression imaging. The present invention introduces snapshot compression sensing into coherent light illumination scenes and applies the snapshot compression sensing theory in coherent light illumination scenarios, achieving high-dimensional data capture with two-dimensional measurements, improving acquisition speed and reducing physical sensor requirements. Simulation results demonstrate that the present invention achieves superior results across various metrics for the reconstruction of self-sparse 3D space-light fields.

Description

Technical Field

[0001] The present invention relates to the field of high-dimensional signal processing, and in particular to a coherent snapshot compression imaging method for high-dimensional signals. Background Art

[0002] The acquisition of high-dimensional data is a long-standing challenge in signal processing and related fields. Most physical optics, X-ray, radar, and acoustic sensors are composed of 2D arrays, but sensors that directly perform 3D (or higher-dimensional) data acquisition generally do not exist. To address this problem, scholars have proposed a snapshot compressed imaging (SCI) method that maps multiple frames into a single measurement frame, uses a 2D detector to capture 3D data, and then uses a reconstruction algorithm to recover the original data.

[0003] Due to its unique sensing matrix, a compressed imaging system can capture raw three-dimensional data using two-dimensional measurements. This imaging system overcomes the difficulties faced by traditional methods in acquiring high-dimensional signals due to hardware limitations, reduces the requirements for physical sensors, and offers the advantages of low cost and low power consumption. Snapshot compressed sensing, due to its flexibility in processing high-dimensional signals, has already found outstanding applications in video and hyperspectral signals, such as the coded aperture compressed temporal imaging system developed for video signals and the single-dispersor coded aperture compressed spectral imaging system for hyperspectral signals.

[0004] Therefore, snapshot compressed sensing has become an important research direction in the field of multidimensional signals. Summary of the Invention

[0005] The present invention proposes a coherent snapshot compression imaging method for high-dimensional signals, which can solve at least one of the above technical problems.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A coherent snapshot compression imaging method for high-dimensional signals comprises the following steps:

[0008] S1. For the space-time optical field propagation of a coherent light illumination scene, the space-time optical field is modulated by a composite phase mask generated by a spatial light modulator;

[0009] S2, superimposing the recording on the detector plane to realize a single hologram produced by space-light field compression modulation obtained by a single exposure;

[0010] S3. Send the hologram to the receiving end and use the existing reconstruction algorithm to reconstruct the original signal to achieve coherent snapshot compression imaging.

[0011] Further, taking 3D as an example, without considering the optical details, the following steps are included:

[0012] SCI uses a unified mathematical model to Representing a 3D data cube using a mask The data cube is modulated and, with represents the kth frame of the data cube, where Likewise, H k represents the corresponding k-th mask; the two-dimensional measurement G is modeled as:

[0013]

[0014] Here N represents the measurement noise;

[0015]

[0016]

[0017] Where f k =vec(F k ) represents the vectorization of the k-th frame by column superposition, D k =Diag(vec(H k )) is a diagonal matrix with diagonal elements H k The vectorized form of , the forward model is as follows:

[0018] g=Hf+n, (4)

[0019] Where g = vec(G), n = vec(N);

[0020] The Gabor hologram g(x,y) recorded within Δt time is expressed as the integral of the light field intensity I(x,y,t) on the detector plane

[0021]

[0022] Said S1 specifically includes:

[0023] First, a beam splitter is used to split the expanded and collimated laser beam into an object beam and a reference beam. The object beam is first irradiated on a 3D space-time light field with a scattering intensity of O(x, y, t), and then propagates to the spatial light modulator plane with a distance Δz1 from the light field. After being modulated by the time-varying mask M(x, y, t), it is parallel to the reference light adjusted by the beam splitter and then propagates a distance Δz2 to the detector plane for recording. Therefore, the light field intensity I(x, y, t) on the detector plane is expressed as shown in formula (6):

[0024]

[0025] Among them, R is the reference light, O(x,y,t) is the object light, P λ,Δz () is a propagation operator with wavelength λ and propagation distance Δz in the spatial domain.

[0026] Furthermore, in free space, the diffraction field propagation is described using the angular spectrum method, as shown in Equation (7).

[0027]

[0028] Among them, the optical transfer function It gives a complex-valued matrix, the real part and imaginary part correspond to the amplitude and phase information of the function in the frequency domain respectively, exp() refers to the exponential function with the natural constant e as the base, i is the imaginary unit, and (u,v) is the spatial frequency component;

[0029] The time-varying mask M(x,y,t) consists of two parts. The first is the time-varying random phase mask used for modulation. The other part is the time-varying lens phase factor Right now in, is a time-varying random matrix uniformly distributed between 0 and 1, is the time-varying lens phase factor;

[0030] Since the time-varying mask is transformed in a discrete form, if the space-time-light field is divided into L frames within the ΔT time, the time-varying mask also changes L times within the ΔT time; assuming that the horizontal sampling interval is set to Δ k , according to the paraxial approximation theory, formula (6) can be rewritten as:

[0031]

[0032] Where: R l With E l They are the reference light field and object light field of the lth frame of space-light field propagated to the detector plane;

[0033] The reference light R is a constant, and the single-frame hologram acquired by the detector is written as:

[0034]

[0035] Furthermore, step S3 includes:

[0036] The hologram recording process within ΔT can be modeled as a snapshot compressed sensing process, defined as:

[0037] {vec[O(n1Δx,n2Δy,l)]} T =f l , Formula (8) can be further expressed as:

[0038]

[0039] in: It represents the sensing matrix corresponding to the space-light field of the first frame;

[0040] Formula (10) can be written as:

[0041] g=Hf+n, (11)

[0042] in: is the forward imaging model within Δt time,

[0043] H=[H1 H2 ... H L ], (12)

[0044]

[0045] Among them: diag() represents a diagonal matrix;

[0046] An estimate of the original 3D space-time light field can be obtained by solving an unconstrained optimization problem:

[0047]

[0048] On the other hand, the present invention discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0049] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0050] It can be seen from the above technical solution that the coherent snapshot compression imaging method for high-dimensional signals of the present invention has the following beneficial effects:

[0051] Because snapshot compressed imaging can encode image sequences into single-frame measurements, it has advantages such as low physical sensor requirements, low bandwidth / memory requirements, and fast acquisition speeds. However, existing snapshot compressed imaging primarily targets incoherent light scenarios, such as video and hyperspectral signals, with limited research on coherent light. This paper introduces snapshot compressed sensing to coherent light illumination scenarios, applying snapshot compressed sensing theory in these scenarios to capture high-dimensional data with two-dimensional measurements, improving acquisition speed and reducing physical sensor requirements. Simulation results demonstrate that the proposed algorithm achieves superior results across various metrics for the reconstruction of self-sparse 3D space-time light fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of time-space coherent snapshot compression imaging according to an embodiment of the present invention;

[0053] Figure 2 Schematic diagram of the SCI encoding and decoding process according to an embodiment of the present invention;

[0054] Figure 3 This is a 3D space-light field compression holographic modulation optical model according to an embodiment of the present invention;

[0055] Figure 4 This is a schematic diagram of 3D space-light field compression holographic modulation and decoding reconstruction recovery

[0056] Figure 5 This is a simulation experiment of self-sparse 3D space-time light field compression imaging of an example of the present invention: (a) frame number simulation experiment; (b) 8-frame self-sparse 3D space-time light field. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0058] The current SCI work is mainly based on incoherent illumination. The coherent snapshot compression imaging method for high-dimensional signals described in the embodiment of the present invention proposes a coherent snapshot compression imaging method for applying SCI to coherent light illumination scenarios.

[0059] In order to realize the above method, the present invention adopts the following technical scheme (such as Figure 1 shown):

[0060] S1. For the space-time optical field propagation of a coherent light illumination scene, the space-time optical field is modulated by a composite phase mask generated by a spatial light modulator;

[0061] S2, superimposing the recording on the detector plane to realize a single hologram produced by space-light field compression modulation obtained by a single exposure;

[0062] S3. Send the hologram to the receiving end and use the existing reconstruction algorithm to reconstruct the original signal.

[0063] Note: The composite phase mask generated by the spatial light modulator in step S1 consists of a random phase mask and a lens phase factor.

[0064] The basic principle of the embodiment of the present invention is still consistent with the traditional SCI, which mainly compresses the 3D data cube into 2D measurement values, and then reconstructs the original 3D data cube from the 2D measurement values using a recovery algorithm, that is, the process of compression encoding and reconstruction decoding. Through these two steps, an optical-hardware-encoder-software-decoder SCI system is obtained. The principle diagram is shown in FIG. Figure 2 shown.

[0065] The following examples illustrate:

[0066] In the description of the embodiment of the present invention, a unified mathematical model is used for SCI (taking 3D as an example without considering optical details). Representing a 3D data cube, using a mask H∈ Nx×N y ×Nt The data cube is modulated and, with represents the kth frame of the data cube, where Likewise, H k Denotes the corresponding k-th mask. The two-dimensional measurement G can be modeled as:

[0067]

[0068] Here N represents the measurement noise.

[0069]

[0070]

[0071] Where f k =vec(F k ) represents the vectorization of the k-th frame (by column stacking), D k =Diag(vec(H k )) is a diagonal matrix with diagonal elements H k The vectorized form of , the forward model is as follows:

[0072] g=Hf+n, (4)where g=vec(G), n=vec(N).

[0073] The Gabor hologram g(x,y) recorded within Δt can be expressed as the integral of the light field intensity I(x,y,t) on the detector plane:

[0074]

[0075] Since the 3D space-light field compression holographic encryption optical experimental device in the present invention is as follows Figure 3As shown in the figure, the expanded and collimated laser beam is firstly divided into object light and reference light by a beam splitter. The object beam is first irradiated on a 3D space-time light field with a scattering intensity of O(x, y, t), and then propagates to the spatial light modulator plane with a distance Δz1 from the light field. After being modulated by the time-varying mask M(x, y, t), it is parallel to the reference light adjusted by the beam splitter and then propagates a distance Δz2 to the detector plane for recording. Therefore, the light field intensity I(x, y, t) on the detector plane is expressed as shown in formula (6).

[0076]

[0077] Among them, R is the reference light, O(x,y,t) is the object light, P λ,Δz () is a propagation operator with wavelength λ and propagation distance Δz in the spatial domain.

[0078] In free space, the diffraction field propagation can be described using the angular spectrum method, as shown in Equation (7).

[0079]

[0080] Among them, the optical transfer function It gives a complex-valued matrix, where the real and imaginary parts correspond to the amplitude and phase information of the function in the frequency domain, respectively. exp() refers to the exponential function with the natural constant e as the base, i is the imaginary unit, and (u,v) is the spatial frequency component.

[0081] The time-varying mask M(x,y,t) consists of two parts. The first is the time-varying random phase mask used for modulation. The other part is the time-varying lens phase factor Right now in, is a time-varying random matrix uniformly distributed between 0 and 1, is the time-varying lens phase factor.

[0082] Since the time-varying mask is transformed in a discrete form, if the space-light field within ΔT time is divided into L frames, the time-varying mask also changes L times within ΔT time. Assume that the horizontal sampling interval is set to Δ k , according to the paraxial approximation theory, formula (6) can be rewritten as:

[0083]

[0084] Where: R l With E l are the reference light field and object light field of the lth frame of space-time light field propagated to the detector plane.

[0085] The reference light R is a constant, and the single-frame hologram acquired by the detector can be written as:

[0086]

[0087] Combine Figure 2 And the relevant SCI statement text, the recording process of the hologram within the ΔT time in the present invention can be modeled as a snapshot compressed sensing process, defined as: {vec[O(n1Δx,n2Δy,l)]} T =f l , Formula (8) can be further expressed as:

[0088]

[0089] in: represents the sensing matrix corresponding to the space-light field of the first frame. Equation (10) can be written as:

[0090] g=Hf+n, (11) in: is the forward imaging model within Δt time,

[0091] H=[H1 H2 … H L ], (12)

[0092]

[0093] Where: diag() represents a diagonal matrix.

[0094] An estimate of the original 3D space-time light field can be obtained by solving an unconstrained optimization problem:

[0095]

[0096] Schematic diagram of 3D space-time light field coherence snapshot compression holographic imaging Figure 4 As shown in Figure 2, a parallel light beam illuminates a space-time optical field (O(x, y, t)) discretized into L frames. It then propagates through a spatial distance Δz1 to the mask plane. After being encoded by a time-varying composite mask composed of a random phase mask and a variable lens phase, it continues to propagate a distance Δz2 to the detector plane. A single exposure captures the entire hologram. During reconstruction, the resulting hologram, the propagation distance, and the time-varying composite mask are combined to reconstruct the original 3D space-time optical field using algorithms such as TwIST.

[0097] The following examples illustrate the effects of the present invention:

[0098] Here we show the performance comparison of the proposed algorithm for the reconstruction of self-sparse 3D space-time light field and the traditional BP reconstruction algorithm.

[0099] The effect of the number of frames to be processed from the sparse 3D space-light field on the compression imaging reconstruction performance of the method proposed in the present invention is tested, such as Figure 5 As shown in (a), as the number of space-light field frames increases, the proposed method slightly reduces the PSNR in the reconstructed space-light field, but still achieves good reconstruction performance, with significant advantages over the BP algorithm. When the self-sparse space-light field to be processed is 8 frames, the reconstructed PSNR is about 42.6dB. When the light field is increased to 16 frames, the PSNR of the compressed imaging reconstruction is about 28.0dB. To show a more obvious visual effect, Figure 5 (b) shows that when the space-light field is 8 frames, within a certain range, the method proposed in the present invention has a good reconstruction effect for the self-sparse 3D space-light field, and the reconstructed space-light field has low distortion and no obvious visual difference.

[0100] In summary, snapshot compression imaging can encode image sequences into single-frame measurements, offering advantages such as low physical sensor requirements, low bandwidth / memory requirements, and fast acquisition speeds. However, existing snapshot compression imaging mainly targets incoherent light scenarios such as video signals and hyperspectral signals, with relatively little research on coherent light.

[0101] This paper introduces snapshot compressed sensing into coherent light illumination scenarios, proposing a coherent snapshot compressed imaging method suitable for 3D space-time light fields. Simulation results show that the proposed algorithm achieves good results across various metrics when reconstructing self-sparse 3D space-time light fields.

[0102] The specific experimental setup of the present invention is as follows. The laser is expanded into reference light and object light by a beam-splitting and collimating optical device group. After the object beam is irradiated on the space-light field of O(x, y, t) discretized into L frames, it propagates through space with a distance of Δz1 to the mask plane. After being encoded by a time-varying composite mask composed of a random phase mask and a variable lens phase, it continues to propagate a distance of Δz2 to the detector plane. The entire hologram can be recorded through a single exposure; the reference light is directly irradiated in space and propagated to the mask plane. After being encoded by a time-varying composite mask composed of a random phase mask and a variable lens phase, it continues to propagate a distance to the detector plane. During reconstruction, the original 3D space-light field can be reconstructed by combining the obtained hologram, the propagation distance and the time-varying composite mask using algorithms such as TwIST.

[0103] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of any of the above methods.

[0104] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of any of the above methods.

[0105] In another embodiment provided by the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute the steps of any one of the methods in the above embodiments.

[0106] It is understandable that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts of the above method.

[0107] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0108] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0109] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A coherent snapshot compression imaging method for high-dimensional signals, characterized in that: The following steps are included: S1. For the space-time optical field propagation of coherent light illumination scenes, the space-time optical field is modulated by a composite phase mask generated by a spatial light modulator; S2, superimposing the recording on the detector plane to realize a single hologram produced by space-light field compression modulation obtained by a single exposure; S3, sending the hologram to the receiving end, and using the reconstruction algorithm to reconstruct the original signal to achieve coherent snapshot compression imaging; In a 3D scene, without considering the optical details, the following steps are included: SCI uses a unified mathematical model to Representing a 3D data cube using a mask The data cube is modulated and, with represents the kth frame of the data cube, where ,same, Represents the corresponding k-th mask; two-dimensional measurement Modeled as: (1) here represents the measurement noise; (2) (3) In the formula represents the vectorization of the k-th frame by column superposition, is a diagonal matrix with diagonal elements The vectorized form of , the forward model is as follows: (4) in , ; Gabor hologram recorded in time Expressed as the light field intensity on the detector plane Points: (5)。 2. The coherent snapshot compression imaging method for high-dimensional signals according to claim 1, characterized in that: Said S1 specifically includes: First, the beam splitter is used to split the expanded and collimated laser beam into object light and reference light. The object beam is first irradiated on a laser beam with a scattering intensity of 3D space-time light field, and then propagates to the distance from the light field The spatial light modulator plane, through the time-varying mask After modulation, it is parallel to the reference light adjusted by the beam splitter and then propagates The distance to the detector plane is recorded, so the light field intensity at the detector plane It is expressed as shown in formula (6), (6) in, For reference light, For the light of things, The wavelength in the spatial domain is , the propagation distance is The propagation operator.

3. The coherent snapshot compression imaging method for high-dimensional signals according to claim 2, characterized in that: The step S2 specifically includes: In free space, the diffraction field propagation is described using the angular spectrum method, as shown in equation (7): (7) Among them, the optical transfer function , which gives a complex-valued matrix, the real and imaginary parts correspond to the amplitude and phase information of the function in the frequency domain, respectively. It refers to the exponential function with the natural constant e as the base. is the imaginary unit, is the spatial frequency component; Time-varying mask It consists of two parts. The first is the time-varying random phase mask used for modulation. , and the other part is the time-varying lens phase factor ,Right now ,in, is a time-varying random matrix uniformly distributed between 0 and 1, is the time-varying lens phase factor; Since the time-varying mask is transformed in discrete form, if The time-space-light field is divided into Frame, time-varying mask in The same changes in time times; assuming the horizontal sampling interval is set to , according to the paraxial approximation theory, formula (6) can be rewritten as: (8) in: and Respectively Frame space - the reference light field and object light field propagated to the detector plane; Reference light is a constant, and the single-frame hologram acquired by the detector is written as: (9)。 4. The coherent snapshot compression imaging method for high-dimensional signals according to claim 3, characterized in that: The step S3 comprises: The temporal hologram recording process can be modeled as a snapshot compressed sensing process, defined as: Formula (8) can be further expressed as: （10） in , indicating the Frame space-the sensor matrix corresponding to the optical field; Formula (10) can be written as: (11) in: , for Forward imaging model in time, (12) (13) in: represents a diagonal matrix; An estimate of the original 3D space-time light field can be obtained by solving an unconstrained optimization problem: (14)。 5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 4.

6. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 4.

Images