A polarization spectrum video compression acquisition system

Abstract

The present invention relates to a polarization spectrum video compression acquisition system, which belongs to the field of computational imaging; it includes a lens group, a spatial coding random modulation module S1, a spectral dispersion modulation module S2, a pixel polarization detector module S3 and a high-precision polarization spectrum data reconstruction module S4; the present invention can realize efficient compression acquisition and reconstruction of multi-dimensional light field data containing polarization information, spectral information and spatial information. Polarization spectrum information in a scene can be compressed and acquired by a single exposure, polarization spectrum information of 4 polarization angles can be acquired by a high-precision compressed sensing reconstruction algorithm, and polarization spectrum video acquisition can be realized in a continuous sampling mode. The present invention can reduce the complexity of system hardware, and has the characteristics of low data acquisition volume and efficient acquisition of high-resolution polarization spectrum video.

Description

Technical Field

[0001] The invention belongs to the field of computational imaging and relates to a polarization spectrum video compression acquisition system. Background Art

[0002] Computational imaging technology integrates knowledge from multiple disciplines such as optics, graphics and information processing, and is a hot topic in the current interdisciplinary field. Through the innovation of imaging modes and the improvement of algorithm processing, computational imaging technology simplifies traditional hardware imaging systems and improves imaging capabilities. In particular, it provides a feasible solution for the simultaneous acquisition of multi-dimensional information, and has broad application prospects in the fields of aerospace remote sensing, security monitoring and special condition imaging.

[0003] Light field information is information with seven-dimensional characteristics. The current detector is a two-dimensional imaging mode, which often achieves the collection of limited-dimensional information by discarding other dimensions. Multi-dimensional information covers more characteristic information of the target. For example, polarization information can reflect relevant information about the target surface, which can be used to detect the particle size, morphology and roughness of the material surface, and can also detect and identify dim targets. In terms of application, it can be used for polarization defogging, atmospheric detection and deep space detection. Spectral information can reflect the component information of the target surface and has target-specific identification features. The simultaneous acquisition of polarization information and spectral information can greatly improve the dimension of information obtained and obtain more comprehensive target information.

[0004] Traditional polarization spectroscopy technology has problems such as noise sensitivity, channel crosstalk, and low spectral resolution, which affect the accuracy of the polarization spectrum obtained. Due to the high-dimensional characteristics of polarization spectroscopy information, it is often necessary to achieve dimensionality reduction acquisition through dimensional segmentation, which sacrifices time resolution, spatial resolution, etc., and cannot achieve the acquisition of polarization spectroscopy video. Computational imaging technology is based on the principle of compressed sensing, which can achieve the reduction of high-dimensional information to acquisition, and then through the coordinate basis projection that meets the sparse projection conditions, the optimization reconstruction algorithm can realize the acquisition of high-dimensional light field information. The development of computational imaging technology has made it possible to obtain high-quality polarization spectroscopy videos. Current computational imaging technology also faces problems such as reconstruction accuracy and low light flux. Therefore, the synchronous acquisition of multi-dimensional information based on computational imaging technology has important significance and broad application prospects. Summary of the invention

[0005] The technical problem solved by the present invention is: to overcome the shortcomings of the prior art and propose a polarization spectrum video compression acquisition system, which can realize the real-time acquisition of polarization, spectrum and spatial multi-dimensional video information through snapshot calculation of the polarization spectrum video acquisition system.

[0006] The solution of the present invention is:

[0007] A polarization spectrum video compression acquisition system, comprising a lens group, a spatial coding random modulation module S1, a spectral dispersion modulation module S2, a pixel polarization detector module S3 and a high-precision polarization spectrum data reconstruction module S4;

[0008] The spatial coding random modulation module S1 receives the light passing through the lens group, and performs random spatial modulation of the light to form coded modulated light, and sends the coded modulated light to the spectral dispersion modulation module S2;

[0009] Spectral dispersion modulation module S2: receives the coded modulated light from the spatial coded random modulation module S1, performs dispersion processing on the coded modulated light using a nonlinear spectral dispersion function, generates the desired coded spectral dispersion light, and sends the desired coded spectral dispersion light to the pixel polarization detector module S3;

[0010] Pixel polarization detector module S3: receives the desired coded spectral dispersion light from the spectral dispersion modulation module S2, extracts polarization information of the desired coded spectral dispersion light, performs photoelectric conversion processing on the light, and then forms an image to generate an aliased blurred image, and sends the aliased blurred image to the high-precision polarization spectral data reconstruction module S4;

[0011] High-precision polarization spectral data reconstruction module S4: receives the aliased blurred image from the pixel polarization detector module S3, pre-processes the aliased blurred image in sequence to realize the separation of aliased polarization information, establishes the objective function and performs high-precision demodulation and reconstruction, combines the forward model and the convex optimization solution algorithm based on the total variation regularization constraint to realize the denoising solution of the polarization spectral information, generates a clear spectral polarization image of the target scene and outputs it.

[0012] In the above-mentioned polarization spectrum video acquisition device and method, the random spatial modulation of the spatial coding random modulation module S1 is:

[0013] Square codes are randomly set on the board of the spatial coding random modulation module S1, and the square codes are numbered 0 or 1; when the square code is numbered 0, the code is in a sealed state and light cannot pass through; when the square code is numbered 1, the code is in a through-hole state and light can pass through.

[0014] In the above-mentioned polarization spectrum video acquisition device and method, the square codes numbered 0 and the square codes numbered 1 each account for 50%; and the square codes numbered 0 are made by chrome plating process.

[0015] In the above-mentioned polarization spectrum video acquisition device and method, the spatial coding random modulation module S1 is located at the primary image plane of the polarization spectrum video compression acquisition system.

[0016] In the above-mentioned polarization spectrum video acquisition device and method, the spectral dispersion modulation module S2 adopts a double Amici dispersion prism; and the dispersion direction of the spectral dispersion modulation module S2 is horizontal.

[0017] In the above-mentioned polarization spectrum video acquisition device and method, the pixel polarization detector module S3 is provided with a detector including a pixel-level lens array and a linear polarizer.

[0018] In the above-mentioned polarization spectrum video acquisition device and method, the objective function established by the high-precision polarization spectrum data reconstruction module S4 is:

[0019]

[0020] Where, F is the original light data;

[0021] X(F) is the objective function of F;

[0022] H is the system equivalent observation matrix;

[0023] Γ(F) is the regularity constraint term;

[0024] τ is the equilibrium parameter of Γ(F);

[0025] High-precision demodulation and reconstruction are achieved according to the objective function.

[0026] In the above-mentioned polarization spectrum video acquisition device and method, a forward model is established in the high-precision polarization spectrum data reconstruction module S4:

[0027]

[0028] Where m is the horizontal coordinate of the two-dimensional plane;

[0029] n is the vertical coordinate of the two-dimensional plane;

[0030] f(m,n) is the data obtained in the forward direction by polarization spectrum compression sampling;

[0031] e is the polarization angle;

[0032] w is the number of spectrum segments;

[0033] λ is the wavelength;

[0034] k is the square code number, k = 0 or 1;

[0035] ψ e is the original input data;

[0036] T is spatially coded random modulation and spectral dispersion modulation;

[0037] h is the subsequent extraction of polarization information;

[0038] η is the noise error introduced when the system is sampled.

[0039] In the above-mentioned polarization spectrum video acquisition device and method, the forward model is a discretization model of continuous sampling.

[0040] In the above-mentioned polarization spectrum video acquisition device and method, when the high-precision polarization spectrum data reconstruction module S4 realizes the denoising solution of polarization spectrum information based on the convex optimization solution algorithm of total variation regularization constraint, the total variation regularization constraint is used to perform two-step iterative shrinkage threshold to complete the denoising solution.

[0041] The beneficial effects of the present invention compared with the prior art are:

[0042] (1) The present invention comprehensively considers the temporal resolution of polarization spectrum data acquisition and solves the problem of effective acquisition of polarization spectrum video;

[0043] (2) The present invention constructs a snapshot computational polarization spectrum imaging model to solve the problem of high-precision reconstruction of aliased blurred images;

[0044] (3) The present invention takes into account the snapshot imaging mode and constructs an imaging acquisition model, which is more suitable for practical applications and has stronger applicability, and solves the application problem of polarization spectrum video acquisition. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the polarization spectrum video compression acquisition system of the present invention;

[0046] Figure 2 This is a schematic diagram of the square coding number 0 of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the embodiments.

[0048] The present invention provides a polarization spectrum video compression acquisition system. In view of the current problems of low efficiency and complex system of polarization spectrum video acquisition, the present invention can realize efficient compression acquisition and reconstruction of multi-dimensional light field data containing polarization information, spectral information and spatial information. Polarization spectrum information in the scene can be compressed and acquired with a single exposure. Polarization spectrum information of four polarization angles can be acquired through a high-precision compressed sensing reconstruction algorithm. Polarization spectrum video acquisition can be realized in a continuous sampling mode. The present invention can reduce the complexity of system hardware, and has the characteristics of low data acquisition volume and efficient acquisition of high-resolution polarization spectrum video.

[0049] Polarization spectrum video compression acquisition system, such as Figure 1 As shown, it specifically includes a lens group, a spatial coding random modulation module S1, a spectral dispersion modulation module S2, a pixel polarization detector module S3 and a high-precision polarization spectrum data reconstruction module S4;

[0050] Spatial coding random modulation module S1: Spatial coding random modulation module S1 is located at the primary image plane of the polarization spectrum video compression acquisition system. It receives the light passing through the lens group, and performs random spatial modulation of the light to form coded modulated light, and sends the coded modulated light to the spectral dispersion modulation module S2.

[0051] The random spatial modulation of the spatial coding random modulation module S1 is:

[0052] Square codes are randomly set on the board of the spatial coding random modulation module S1, and the square codes are numbered 0 or 1; when the square code number is 0, it means that the code is in a sealed state and light cannot pass through, such as Figure 2 As shown. When the square code number is 1, the code is in a through-hole state, and light can pass through. The square code numbered 0 and the square code numbered 1 each account for 50%; and the square code numbered 0 is made using a chrome plating process. The random spatial modulation of the spatial coding random modulation module S1 is a random coding (0 / 1) coding matrix, which realizes the modulation of not passing or passing the spatial light.

[0053] Spectral dispersion modulation module S2: receives the coded modulated light from the spatially coded random modulation module S1, performs dispersion processing on the coded modulated light using a nonlinear spectral dispersion function, generates the desired coded spectral dispersion light, and sends the desired coded spectral dispersion light to the pixel polarization detector module S3.

[0054] The spectral dispersion modulation module S2 adopts a double Amici dispersion prism and a nonlinear dispersion model, and realizes dispersion modulation in the spectral dimension after being relayed by a relay mirror after the spatial coding modulation module described in S1. And the dispersion direction of the spectral dispersion modulation module S2 is horizontal. The spectral dispersion modulation module S2 realizes the dispersion of the light after spatial coding modulation into spectra of different wavelengths.

[0055] The spectral dispersion modulation module S2 is located between the relay lens group and the detector, and realizes the calibration data collection of the dispersion unit through sampling and calibration of different wavelengths of the monochromator.

[0056] Pixel polarization detector module S3: receives the desired coded spectral dispersion light from the spectral dispersion modulation module S2, extracts polarization information of the desired coded spectral dispersion light, performs photoelectric conversion processing on the light, and then forms an image to generate an aliased blurred image, and sends the aliased blurred image to the high-precision polarization spectral data reconstruction module S4.

[0057] The pixel polarization detector module S3 is provided with a detector of a pixel-level lens array and a linear polarizer. The pixel polarization detector module S3 realizes the collection of two-dimensional compressed polarization spectrum aliasing images.

[0058] The pixel polarization detector module S3 is plated with polarizers with different polarization angles to achieve polarization information modulation at different angles.

[0059] High-precision polarization spectral data reconstruction module S4: receives the aliased blurred image from the pixel polarization detector module S3, pre-processes the aliased blurred image in sequence to realize the separation of aliased polarization information, establishes the objective function and performs high-precision demodulation and reconstruction, combines the forward model and the convex optimization solution algorithm based on the total variation regularization constraint to realize the denoising solution of the polarization spectral information, generates a clear spectral polarization image of the target scene and outputs it.

[0060] The objective function established by the high-precision polarization spectrum data reconstruction module S4 is:

[0061]

[0062] Where, F is the original light data;

[0063] X(F) is the objective function of F;

[0064] H is the system equivalent observation matrix;

[0065] Γ(F) is the regularity constraint term;

[0066] τ is the equilibrium parameter of Γ(F);

[0067] High-precision demodulation and reconstruction are achieved according to the objective function.

[0068] A forward model is established in the high-precision polarization spectrum data reconstruction module S4:

[0069]

[0070] Where m is the horizontal coordinate of the two-dimensional plane;

[0071] n is the vertical coordinate of the two-dimensional plane;

[0072] f(m,n) is the data obtained in the forward direction by polarization spectrum compression sampling;

[0073] e is the polarization angle;

[0074] w is the number of spectrum segments;

[0075] λ is the wavelength;

[0076] k is the square code number, k = 0 or 1;

[0077] ψ e is the original input data;

[0078] T is spatially coded random modulation and spectral dispersion modulation;

[0079] h is the subsequent extraction of polarization information;

[0080] η is the noise error introduced when the system is sampled.

[0081] The forward model is a discretized model for continuous sampling.

[0082] When the high-precision polarization spectral data reconstruction module S4 realizes the denoising solution of polarization spectral information based on the convex optimization solution algorithm of total variation regularization constraint, the total variation regularization constraint is used to perform two-step iterative shrinkage threshold to complete the denoising solution.

[0083] Example

[0084] Polarization spectrum video compression acquisition system

[0085] It includes: a spatial coding random modulation module S1, a spectral dispersion modulation module S2, a pixel polarization detector module S3, and a high-precision polarization spectrum data reconstruction module S4. Among them, the S1 spatial coding is a randomly coded (0 / 1) coding matrix to realize the modulation of spatial light not passing through or passing through. The S2 spectral dispersion module adopts a double Amici prism and a nonlinear dispersion model, and realizes the dispersion modulation of the spectral dimension after the spatial coding modulation module in S1 and the relay mirror. The S2 spectral dispersion modulation module disperses the light after the spatial coding modulation in S1 into spectra of different wavelengths. The S2 spectral dispersion modulation module is located between the relay lens group and the detector, and the calibration data collection of the dispersion unit is realized by sampling and calibrating different wavelengths of the monochromator. The S3 pixel polarization detector module is located after the spectral dispersion modulation module in step S2. The S3 pixel polarization detector realizes the collection of two-dimensional compressed polarization spectrum aliasing images. The S3 pixel polarization detector is plated with polarizers with different polarization angles to realize polarization information modulation at different angles. The S4 high-precision polarization spectrum data reconstruction module receives the aliased blurred image, preprocesses the aliased blurred image to separate the aliased polarization information, performs high-precision demodulation and reconstruction, and then obtains and outputs a clear spectral polarization image of the target scene.

[0086] When designing the spatial coding random modulation module S1, a transmittance of 50% is used to ensure the light flux provided by the polarization spectrum acquisition system. The size of the spatial coding random modulation module S1 needs to be designed to match the focal length and the camera pixel size. The densely etched codes of 1 (pass) or 0 (not pass) on the spatial coding random modulation module S1 are used to achieve spatial modulation of the scene. In one embodiment of the present invention, an aperture size of 6.9μm×6.9μm is selected.

[0087] After passing through the spatial coding random modulation module S1, the light enters the rear spectral dispersion modulation module S2, which then disperses the spatially modulated light into spectra of different wavelengths and forms an image on the rear pixel polarization detector module S3.

[0088] In one embodiment of the present invention, the spectral dispersion modulation module uses a double Amici prism, which allows the light of the central wavelength to be emitted from the last plane at an angle parallel to the incident light, and the light of other wavelengths is emitted according to the refractive index of the glass material, ultimately achieving spectral dispersion modulation of the incident light.

[0089] The S3 pixel polarization detector module is a detector with pixel-level tiny polarizers plated on the surface of the pixel, which can acquire polarized light at 0°, 45°, 90° and 135°, and realize polarization splitting.

[0090] The polarization spectrum compression sampling forward model of the high-precision polarization spectrum data reconstruction module S4 is constructed according to the above imaging process, and the spatial coding modulation and spectral dispersion modulation of the polarization spectrum are realized through the data model.

[0091] In the embodiment of the present invention, the polarization spectrum compressive sampling forward model can be expressed as:

[0092]

[0093] f represents the data obtained in the embodiment, ψ e is the input raw data; T is the spatial coding random modulation and spectral dispersion modulation; h is the subsequent extraction of polarization information; η is the noise error introduced during system sampling.

[0094] In the embodiment of the present invention, LSH-T50 is used as the standard light source, and Zolix Omni-λ300 monochromator is used as the standard output spectrum calibration reference. The spectrum is calibrated every 1 nm from 450 nm to 650 nm, and a suitable spectrum segment is selected as the calibration matrix.

[0095] The data collected by the pixel polarization detector module S3 is combined with the polarization spectrum compression sampling forward model to realize the solution and reconstruction of polarization spectrum data in the high-precision polarization spectrum data reconstruction module S4. Each frame of polarization spectrum image obtained by solution and reconstruction is connected in time sequence to obtain polarization spectrum video.

[0096] The high-precision polarization spectrum data reconstruction module S5 adopts a convex optimization reconstruction model and establishes the following objective equation:

[0097]

[0098] Wherein, G represents the aliased image acquired by the pixel polarization detector module S3, H represents the equivalent sampling function of the system, F represents the input data of the target scene, τ represents the regularization term parameter to adjust the weight of the regularization constraint, and Γ is the regularization constraint term. The target equation is solved by a convex optimization reconstruction algorithm.

[0099] In the embodiment of the present invention, the optimized and improved TwIST algorithm is used to solve the system objective function. In the traditional TwIST algorithm, the total variation denoising is performed on the initial value of the iteration to suppress the decrease in reconstruction accuracy caused by the continuous amplification of noise in subsequent iterative reconstruction.

[0100] In the embodiment of the present invention, a total variation regularization constraint is selected, which is applicable to the condition of smooth spectral transition faced by the present invention.

[0101] According to the polarization spectral video compression sampling system of the embodiment of the present invention, snapshot collection of polarization spectral data can be achieved. The device has no moving parts and a compact structure. High-resolution and high-spectral-precision polarization spectral imaging data can be reconstructed. The snapshot imaging method supports the acquisition of polarization spectral videos.

[0102] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A polarization spectrum video compression acquisition system, characterized by: It includes a lens group, a spatial coding random modulation module S1, a spectral dispersion modulation module S2, a pixel polarization detector module S3 and a high-precision polarization spectrum data reconstruction module S4; The spatial coding random modulation module S1 receives the light passing through the lens group, and performs random spatial modulation of the light to form coded modulated light, and sends the coded modulated light to the spectral dispersion modulation module S2; Spectral dispersion modulation module S2: receives the coded modulated light from the spatial coded random modulation module S1, performs dispersion processing on the coded modulated light using a nonlinear spectral dispersion function, generates the desired coded spectral dispersion light, and sends the desired coded spectral dispersion light to the pixel polarization detector module S3; Pixel polarization detector module S3: receives the desired coded spectral dispersion light from the spectral dispersion modulation module S2, extracts polarization information of the desired coded spectral dispersion light, performs photoelectric conversion processing on the light, and then forms an image to generate an aliased blurred image, and sends the aliased blurred image to the high-precision polarization spectral data reconstruction module S4; High-precision polarization spectrum data reconstruction module S4: receives the aliased blurred image from the pixel polarization detector module S3, pre-processes the aliased blurred image in sequence to split the aliased polarization information, establishes the objective function and performs high-precision demodulation and reconstruction, combines the forward model and the convex optimization solution algorithm based on the total variation regularization constraint to realize the denoising solution of the polarization spectrum information, generates a clear spectral polarization image of the target scene and outputs it; The objective function established by the high-precision polarization spectrum data reconstruction module S4 is: Where, F is the original light data; X(F) is the objective function of F; H is the system equivalent observation matrix; Γ(F) is the regularity constraint term; τ is the equilibrium parameter of Γ(F); Achieve high-precision demodulation and reconstruction according to the objective function; A forward model is established in the high-precision polarization spectrum data reconstruction module S4: Where m is the horizontal coordinate of the two-dimensional plane; n is the vertical coordinate of the two-dimensional plane; f(m,n) is the data obtained in the forward direction by polarization spectrum compression sampling; e is the polarization angle; w is the number of spectrum segments; λ is the wavelength; k is the square code number, k = 0 or 1; ψ e is the original input data; T is spatially coded random modulation and spectral dispersion modulation; h is the subsequent extraction of polarization information; η is the noise error introduced when the system is sampled.

2. The polarization spectrum video compression acquisition system according to claim 1, characterized in that: The random spatial modulation of the spatial coding random modulation module S1 is: Square codes are randomly set on the board of the spatial coding random modulation module S1, and the square codes are numbered 0 or 1; when the square code is numbered 0, the code is in a sealed state and light cannot pass through; when the square code is numbered 1, the code is in a through-hole state and light can pass through.

3. A polarization spectrum video compression acquisition system according to claim 2, characterized in that: The square codes numbered 0 and 1 each account for 50%; and the square codes numbered 0 are made by chrome plating technology.

4. The polarization spectrum video compression acquisition system according to claim 1, characterized in that: The spatial coding random modulation module S1 is located at the primary image plane of the polarization spectrum video compression acquisition system.

5. The polarization spectrum video compression acquisition system according to claim 1, characterized in that: The spectral dispersion modulation module S2 adopts a double Amici dispersion prism; and the dispersion direction of the spectral dispersion modulation module S2 is horizontal.

6. The polarization spectrum video compression acquisition system according to claim 1, characterized in that: The pixel polarization detector module S3 is provided with a detector having a pixel-level lens array and a linear polarizer.

7. The polarization spectrum video compression acquisition system according to claim 1, characterized in that: The forward model is a discretized model for continuous sampling.

8. The polarization spectrum video compression acquisition system according to claim 1, characterized in that: When the high-precision polarization spectrum data reconstruction module S4 implements the denoising solution of polarization spectrum information based on the convex optimization solution algorithm of total variation regularization constraint, the denoising solution is completed by performing two-step iterative shrinkage threshold using total variation regularization constraint.

Images