Polarization defocus plane imaging system calibration method and imaging system

Abstract

The application relates to a calibration method of a polarization defocus plane imaging system and the imaging system, and relates to the fields of photoelectric imaging and remote sensing detection imaging. The application solves the problems of low correction efficiency and difficulty in correcting the system for polarization state changes in the existing polarization error correction technology. The calibration method is based on collected uniform light fields to obtain a plurality of encoding patterns and corresponding polarization modulation matrices as calibration results. The polarization error correction method is based on a target scene to obtain an observation intensity vector; based on the encoding pattern, a corresponding polarization modulation matrix is obtained according to an encoding polarization modulation matrix lookup table, and then a down-sampling Stokes vector after correction of the polarization error is obtained; and the correction of the polarization error is completed through an imaging reconstruction method. A polarization defocus plane imaging system comprises an imaging lens, a DMD (digital micro-mirror array) encoding module, a defocus plane DoFP polarization detector and an imaging reconstruction module. The application is suitable for the fields of remote sensing detection, target identification and environment monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of photoelectric imaging and remote sensing imaging, in particular to the field of polarization imaging. BACKGROUND

[0002] The polarization state of light carries physical information such as target surface texture, roughness, complex refractive index, etc. that cannot be obtained by intensity detection. In the field of remote sensing detection, polarization imaging technology has significant advantages in cloud removal, camouflage recognition, sea surface oil spill monitoring, etc. With the increasing demand for applications, future polarization detection will inevitably develop towards higher spatial resolution and higher radiation / polarization measurement accuracy.

[0003] Division of Focal Plane (DoFP) polarization detectors have become the current mainstream detection method due to their compact structure and real-time imaging characteristics. This detector usually integrates multiple micro-polarizers with different polarization directions in the same focal plane, and can simultaneously obtain intensity images of multiple polarization directions under single exposure conditions, thereby realizing instantaneous polarization information detection.

[0004] The DoFP polarization detector has the advantages of compact structure and instantaneous imaging, but its inherent imaging method also introduces some problems that cannot be ignored:

[0005] (1) Loss of spatial resolution: Since the four micro-polarizers with different polarization directions (0, 45, 90, 135) are arranged in a 2x2 array, the single-channel resolution of the output image is only 1 / 4 of the physical resolution of the detector.

[0006] (2) Instantaneous field of view error: The pixel positions corresponding to different polarization directions are spatially misaligned, which easily introduces field of view inconsistencies between polarization channels, resulting in artifacts in the detection edge or high-frequency texture area, affecting the polarization parameter calculation accuracy.

[0007] Computational optical imaging technology has developed rapidly. This technology combines hardware optical encoding with computational reconstruction algorithms to some extent, breaking through the limitations of traditional imaging systems in terms of resolution, information dimension, and system flexibility.

[0008] Imaging based on spatial light modulators (SLMs) (such as DMDs and liquid crystal spatial light modulators) is one of the important forms of computational imaging. To overcome the physical limitations of traditional optics, computational optical imaging technology based on optical coding modulation has been introduced into polarization-focused plane detection systems. These systems temporally or spatially encode the incident light field and combine this with algorithms for information demodulation and reconstruction. By using hardware optical coding in conjunction with algorithmic reconstruction (such as compressed sensing, super-resolution reconstruction, and spectral reconstruction), they can solve the resolution degradation and instantaneous field-of-view error problems of DoFPs, acquire high-resolution images, and improve imaging quality. However, such computational imaging systems based on polarization-focused planes typically have a large amount of polarization error, leading to decreased imaging contrast and reduced accuracy of the polarization detector.

[0009] The main sources of polarization error are twofold. One is the polarization error introduced by the DoFP imaging method, including: (1) inter-pixel crosstalk and polarization crosstalk: due to the manufacturing process of the micro polarizer and the coupling between pixels, there is crosstalk between different polarization channels, which causes the measured polarization state to deviate from the true value. (2) Extinction ratio and non-uniformity limitation: the extinction ratio of the micro polarizer is usually lower than that of traditional crystal polarization devices, and the non-uniformity of the response of each pixel will directly introduce fixed mode noise, which seriously affects the inversion accuracy of the Stokes vector.

[0010] Second, polarization errors introduced by coded aperture computational imaging include: (1) System complexity error: Adding coded aperture elements and relay optical elements (lenses, prisms, etc.) will introduce additional aberrations and optical path alignment errors. (2) The coding process will change the polarization state of light: Taking the use of digital micromirror array (DMD) as an example, the tilt reflection characteristics of DMD micromirrors will cause polarization-related changes in reflection efficiency and phase difference, thereby changing the polarization state of light under different coding states.

[0011] While existing technologies have made significant progress in correcting polarization errors, especially under static conditions where system state and scene parameters are known, enabling effective correction through accurate modeling and prior calibration, their adaptability remains a serious challenge when faced with dynamic changes in the target, environment, or their own state.

[0012] For example, Chinese patent document CN120820098A discloses a "polarization correction method and system for a linear structured light sensor based on reflected polarized light." This technical solution constructs a complete chain of polarization angle, incident angle, and image processing error control, utilizing a minimum light intensity strategy, incident angle optimization, and differential imaging system integration to reduce polarization errors. This technical solution, starting from prior calibration and static models, establishes an error control chain that highly relies on preset target material and a fixed incident angle range. Therefore, it can only correct static system errors where the scene and state are known, and cannot cope with changes in polarization state caused by dynamic changes in target material or illumination angle during operation.

[0013] Chinese patent document CN116399451A discloses a "Simplified Method for Obtaining Polarization Aberrations Applicable to Planar Symmetric Optical Systems." This technical solution is based on a precise three-dimensional polarization ray tracing method. It directly represents the coefficients of various polarization aberrations using the precise values ​​of the bidirectional attenuation and phase delay of the rays at the upper edge, center, and lower edge. Based on this, the polarization aberration at any coordinate in the pupil is calculated, thereby correcting polarization errors. This technical solution starts from the perspective of fixed-structure optical design, and its polarization aberration model is calculated based on the stationary and unchanging parameters of the system. Therefore, it can only correct static aberrations inherent in the system under the design state, and cannot correct dynamic errors caused by the real-time evolution of optical performance due to temperature, stress, or scanning motion during operation.

[0014] In summary, existing polarization error correction techniques suffer from low correction efficiency and difficulty in correcting systems with changes in polarization state. Summary of the Invention

[0015] This invention alleviates the problems of low correction efficiency and difficulty in correcting systems with changing polarization states in existing polarization error correction techniques. This invention provides the following solution:

[0016] Option 1: A calibration method for a polarization-splitting plane imaging system, wherein the polarization-splitting plane imaging system includes:

[0017] The front-facing imaging lens projects the light field of the target scene onto the DMD digital micromirror array encoding module to form an optical image, obtaining the Stokes vector of the optical image. ;

[0018] The DMD digital micromirror array encoding module can generate A coding pattern, used to employ any one coding pattern For the Stokes vector Encoding and modulation are performed, and the encoded and modulated optical image is projected onto the relay imaging lens. The light field Stokes vector is... , =1, 2, ..., ;

[0019] The relay imaging lens projects the optical image onto the focal plane DoFP polarization detector.

[0020] The focal plane DoFP polarization detector performs exposure imaging and obtains the polarization response vector. ;

[0021] The calibration method includes the following steps:

[0022] The DMD digital micromirror array encoding module sequentially generates one encoding pattern. For each of the coded patterns The target scene is set sequentially as follows: A uniform light field with polarization states is obtained. Group polarization parameters, based on the The polarization parameters are used to obtain the corresponding polarization modulation matrix. ;

[0023] Will Each coded pattern and its corresponding The polarization modulation matrices are combined into an coded polarization modulation matrix lookup table, which serves as the calibration result for the polarization-focusing plane imaging system.

[0024] Furthermore, in one embodiment of the present invention, the... A uniform optical field with various polarization states is generated using a high-precision polarization state generator.

[0025] Furthermore, in one embodiment of the present invention, the... The value is 16, and the polarization states include 4 circular polarization states and 12 linear polarization states.

[0026] Furthermore, in one embodiment of the present invention, the... The value is 12, and the polarization state includes 12 linear polarization states.

[0027] Furthermore, in one embodiment of the present invention, the... Group polarization parameters include Stokes vector and its corresponding polarization response ,

[0028] .

[0029] Furthermore, in one embodiment of the present invention, the exposure imaging of the focal plane DoFP polarization detector includes extracting the intensity response values ​​of the four polarization channels of each superpixel to form a polarization response vector. The polarization angles of the four polarization channels are 0°, 45°, 90° and 135°, respectively.

[0030] Furthermore, in one embodiment of the present invention, the basis of the... The polarization parameters are used to obtain the corresponding polarization modulation matrix. The method is to minimize the error function using the least squares method.

[0031]

[0032] To achieve, among which, polarization response corresponding to group polarization parameters The matrix formed for Stokes vector corresponding to a uniform optical field in a group of polarization states The matrix formed.

[0033] Furthermore, in one embodiment of the invention, the Stokes vector pass Calculated For the first The Stokes vector corresponding to a uniform light field with polarization states.

[0034] Furthermore, in one embodiment of the present invention, the polarization response It is aimed at the first A uniform light field with various polarization states, and the polarization response vector obtained during exposure imaging by a DoFP polarization detector in a focal plane.

[0035] Option 2: A polarization-splitting plane imaging system, comprising a front imaging lens 2, a DMD digital micromirror array encoding module 3, a relay imaging lens 4, a polarization-splitting plane DoFP polarization detector 5, and an imaging reconstruction module.

[0036] The front imaging lens 2 is used to project the light field of the target scene 1 onto the DMD digital micromirror array encoding module to form an optical image, the Stokes vector of the optical image being... ;

[0037] The DMD digital micromirror array encoding module is capable of generating A coding pattern, used to employ any one coding pattern For the Stokes vector Encoding and modulation are performed, and the encoded and modulated optical image is projected onto the relay imaging lens. The Stokes vector of this optical image is... , =1, 2, ..., ;

[0038] The relay imaging lens projects the optical image onto the focal plane DoFP polarization detector;

[0039] The focal plane DoFP polarization detector performs exposure imaging and obtains the polarization response vector. ;

[0040] The imaging reconstruction module is used to reconstruct the image based on the coded pattern. The corresponding polarization modulation matrix is ​​obtained by looking up the coded polarization modulation matrix in the lookup table. ;

[0041] The coded polarization modulation matrix lookup table is obtained using the calibration method for the coded polarization modulation matrix described in Scheme 1;

[0042] Also used to pass

[0043]

[0044] Obtain the Stokes vector after polarization error correction ,in, The polarization modulation matrix The pseudo-inverse matrix;

[0045] Also used based on the Stokes vector The encoded and modulated Stokes vector is obtained through imaging reconstruction methods. Decoding and reconstruction are performed to obtain a polarized image of the target scene with polarization error removed, polarization error correction is completed, and a reconstructed image is obtained.

[0046] The polarization-focusing plane imaging system calibration method and imaging system described in this invention effectively alleviate the problems of low correction efficiency and difficulty in correcting systems with changes in polarization state in existing polarization error correction techniques. Specific beneficial effects include:

[0047] 1. The image correction method in the imaging system described in this invention relies on the coded polarization modulation matrix lookup table obtained by the calibration method described in this invention. This correction method breaks through the limitations of traditional polarization correction, which relies solely on static optical systems (such as fixed lenses or the detector itself). It successfully incorporates the system's dynamically changing polarization errors into a controllable correction framework, achieving real-time, high-precision polarization error compensation for dynamic coding devices. Through the "coded-polarization modulation matrix" lookup table, it can actively perform one-to-one dynamic correction for time-varying polarization effects (such as changes in reflection efficiency and phase difference caused by micromirror tilt) generated when coding devices (such as DMDs) switch between different coding patterns. This effectively eliminates dynamic aberrations introduced by changes in coding state during computational imaging, ensuring the accuracy of polarization data for each frame of coded image.

[0048] 2. The polarization imaging system described in this invention significantly improves the reconstructed image quality and detection accuracy of a focal plane computational imaging system. By constructing a full-link polarization imaging transmission model, inherent detector defects such as inter-pixel crosstalk and micro-polarizer extinction ratio limitations, as well as aberrations in the relay optical path, are uniformly incorporated into the polarization modulation matrix for inverse calculation. This not only restores the true Stokes vector but also provides high-quality data input without polarization errors for subsequent computational imaging algorithms (such as compressed sensing spectral restoration and compressed sensing super-resolution reconstruction). In other words, it avoids the involvement of "error data" in reconstruction, thereby ultimately obtaining a high-resolution polarization image with higher contrast, clearer details, and no artifacts.

[0049] The method described in this invention is applicable to fields requiring high-resolution, high-precision polarization detection, such as remote sensing, target identification, environmental monitoring, and precision optical measurement. Attached Figure Description

[0050] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0051] Figure 1 This is an architectural diagram of the polarization imaging system described in Implementation Method Six;

[0052] Figure 2 This is a flowchart of the DoFP detection imaging described in Implementation Method Six.

[0053] Figure label:

[0054] Target scene 1; front imaging lens 2; DMD digital micromirror array encoding module 3; relay imaging lens 4; focal plane DoFP polarization detector 5. Detailed Implementation

[0055] Various embodiments of the present invention will now be clearly and completely described with reference to the accompanying drawings. The embodiments described with reference to the drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0056] Implementation Method 1: The polarization-focusing plane imaging system calibration method described in this embodiment includes a front imaging lens, a DMD digital micromirror array encoding module, a relay imaging lens, and a focal plane DoFP polarization detector. The DMD digital micromirror array encoding module is capable of generating... A coded pattern.

[0057] The method includes the following steps:

[0058] Step S0: The front imaging lens acquires a uniform light field with a known polarization state;

[0059] Step S1, the DMD digital micromirror array encoding module sequentially generates one encoding pattern. , =1, 2, ..., ;

[0060] Step S2, based on each of the encoded patterns The uniform light field is modulated to obtain the corresponding polarization modulation matrix. ;

[0061] Step S3, will Each coded pattern and its corresponding The polarization modulation matrices are combined into an coded polarization modulation matrix lookup table, which serves as the calibration result for the polarization-focusing plane imaging system.

[0062] In this embodiment, the number of encoded patterns generated by the DMD digital micromirror array encoding module in step S1 The type is determined based on the specific requirements of the image reconstruction task. For super-resolution reconstruction, the value of N must satisfy N ≥ super-resolution magnification^2 (for example, to achieve 2x super-resolution, at least N=4 is required). The specific form of the encoding pattern matrix can be a Gaussian random matrix, a Bernoulli matrix, a Hadamard matrix, or a specific encoding matrix optimized by deep learning.

[0063] This embodiment illustrates the configuration for a 4x super-resolution imaging task: According to compressed sensing and super-resolution reconstruction theory, the number of coded patterns N must satisfy the sampling condition that N ≥ super-resolution ratio^2. To achieve a 4x improvement in spatial resolution (i.e., a super-resolution ratio of 4), this embodiment sets the number of coded patterns N=16 to ensure that sufficient spatial frequency domain information is acquired for high-fidelity image reconstruction.

[0064] In this embodiment, the encoded and modulated Stokes vector ,pass Obtained through calculation. This method represents the encoded pattern. Modulation of Stokes vector by spatial integral Then, obtain the Stokes vector. . Defined as a 'generalized modulation-integral operator', its specific mathematical operation rules are adaptively defined according to the physical mechanism of the imaging task.

[0065] The definition of the physical mapping process varies depending on the task: for example, in polarization spectral imaging, the operation characterizes the gating and integration process of the spectral dimension; while for super-resolution reconstruction, the operation characterizes the spatial downsampling and energy integration mapping process from the high-resolution spatial grid to the low-resolution detector pixels.

[0066] In this embodiment, a high-precision polarization state generator is used to produce a uniform light field with a known polarization state as the input light source. The polarization modulation matrix includes coded aperture polarization effects, lens errors, and DoFP micro-polarization array errors.

[0067] This embodiment describes a calibration method for a polarization-focusing plane imaging system. This method calibrates the polarization error of dynamic encoding, establishes an "encoding-polarization modulation matrix lookup table," and obtains the polarization modulation matrix under different encoding patterns. The polarization modulation matrix It includes polarization effects of the coding aperture, lens errors, and DoFP micro-polarization array errors, and quantitatively characterizes all polarization errors (including dichroic attenuation, phase delay, depolarization effect, and crosstalk) introduced after light passes through the front lens, coding device, relay imaging lens, and DoFP detector.

[0068] Implementation Method Two: This implementation method further defines the polarization-focusing plane imaging system calibration method described in Implementation Method One. In this implementation method, step S2 is:

[0069] Step S21, based on the encoded pattern ,right The polarization states were obtained respectively Group polarization parameters, the first The polarization parameters include the coded modulated Stokes vector. and polarization response ;

[0070] Step S22, based on the The polarization parameters are grouped, and the error function is minimized using the least squares method to obtain the polarization modulation matrix. ;

[0071] The formula for the error function is:

[0072] .

[0073] in, for polarization response corresponding to group polarization parameters The matrix formed for Stokes vector corresponding to a uniform optical field in a group of polarization states The matrix formed.

[0074] In this embodiment, it is preferable to set up and initialize the optical path before step S21, specifically as follows:

[0075] A uniformly polarized light source is placed in front of the system's front imaging lens, ensuring that the light source covers the detector's field of view; the encoding modulation device is put into operation, and the preset encoding pattern sequence {C1, C2, ..., C} is prepared. N}

[0076] In this embodiment, the In polarization states The polarization state is 12 to 16. When the polarization state is 16, it indicates that the polarization state includes 4 circular polarization states and 12 linear polarization states. When =12, it indicates that the polarization states include only 12 linear polarization states. In passive imaging scenarios in nature and most man-made environments, the light reflected or radiated by the target mainly exhibits linear polarization characteristics, and its circular polarization component is usually extremely weak or even close to zero. Therefore, ignoring the circular polarization state will not result in the loss of the main target polarization characteristics. The setting of =12 is an optimized design for target scenarios that mainly exhibit linear polarization characteristics. By abandoning the detection of circular polarization components with low application value, it achieves high redundancy and high signal-to-noise ratio measurement of key linear polarization information, thus ensuring the accuracy of the detection.

[0077] In this embodiment, the polarization response Obtained through focal plane polarization detection, including the following steps:

[0078] Select the k-th coded pattern The data is loaded onto the encoding and modulation device, and the device remains stable. The following data acquisition cycle is then performed:

[0079] (a) Setting the input polarization state: By adjusting the light source settings and calibrating using a standard polarization state measuring instrument, the input polarization state can be generated. A group of polarized beams with different polarization states;

[0080] (b) Acquisition System Response: Control the focal plane polarization detector for exposure imaging, extract the intensity response values ​​of the four polarization channels (0°, 45°, 90°, 135°) of each superpixel, and construct the polarization response. ;

[0081] (c) Traversing polarization states: Change the polarization state of the light source and repeat steps (a) and (b) at least M times (usually ≥4, to solve the overdetermined equations), thereby obtaining multiple sets of input-output mapping data pairs. .

[0082] Implementation method three is a further limitation on the polarization-focusing plane imaging system calibration method described in implementation method two. In this implementation method, the encoded and modulated Stokes vector... ,pass Obtained through calculation.

[0083] In this embodiment, the encoded and modulated Stokes vector , is encoding modulation The result is obtained after spatial integration. Specific mathematical operation rules are adaptively defined based on the physical mechanism of the imaging task. For example, for polarization spectral imaging tasks, this operator represents band-by-band dot product and integration along the spectral dimension, i.e. As a spectral dimension, the incident light S in Filter energy gating is performed. For the super-resolution reconstruction task of interest in this embodiment, this operator represents the mask modulation and downsampling mapping in the spatial dimension, i.e., the high-resolution coded pattern. With high-resolution light field S in After point-by-point action, energy integration and discretization summation are performed within the spatial aperture of the physical pixels of the detector, thereby completing the projection from the high-dimensional signal space to the low-dimensional measurement space.

[0084] Implementation Method Four: This implementation method further defines the polarization-focusing plane imaging system calibration method described in Implementation Method Two. In this implementation method, the polarization response... By modifying the encoded and modulated Stokes vector Obtained by performing focal plane polarization detection.

[0085] Implementation Method 5: The polarization-focusing plane imaging system correction method described in this implementation method includes a front imaging lens, a DMD digital micromirror array encoding module, a relay imaging lens, and a focal plane DoFP polarization detector; the method includes the following steps:

[0086] Step A1, based on the coded pattern The target scene is acquired by the polarization-focusing plane imaging system to obtain the original intensity image of the target scene;

[0087] Step A2: Extract the observed intensity vector from the original intensity image. ;

[0088] Step A3, based on the coded pattern The corresponding polarization modulation matrix is ​​obtained by looking up the coded polarization modulation matrix in the lookup table. ;

[0089] The coded polarization modulation matrix lookup table is obtained by any of the calibration methods for the coded polarization modulation matrix described in embodiments 1 to 4.

[0090] Step A4, through

[0091]

[0092] Obtain the Stokes vector after polarization error correction ,in, The polarization modulation matrix The pseudo-inverse matrix;

[0093] Step A5, based on the Stokes vector after correcting the polarization error By using imaging reconstruction methods, a polarization image of the target scene with polarization error removed is obtained, thus completing the correction of the polarization error.

[0094] In this embodiment, the Stokes vector , usually includes .

[0095] The polarization error correction method described in this embodiment is used in the actual imaging process. It uses an "encoded-polarization modulation matrix lookup table" to correct the polarization error of the acquired image, obtains Stokes information without the influence of polarization error, and uses the Stokes information without the influence of polarization error to perform computational imaging reconstruction.

[0096] Implementation method six: A polarization-focusing plane imaging system as described in this implementation method, such as... Figure 1 As shown, it includes a front imaging lens, a DMD digital micromirror array encoding module, a relay imaging lens, a focal plane DoFP polarization detector, and an imaging reconstruction module.

[0097] The front-facing imaging lens projects the optical information of the target scene onto the DMD digital micromirror array encoding module to obtain an optical image. The optical information includes Stokes vectors. ;

[0098] The DMD digital micromirror array encoding module is used to load the encoding pattern. It is also used based on the coded pattern. For the Stokes vector Encoding and modulation are performed to obtain the encoded and modulated Stokes vector. Thus, the encoded and modulated optical field is obtained;

[0099] The relay imaging lens is used to project the encoded and modulated light field onto the focal plane DoFP polarization detector to obtain a secondary image;

[0100] The focal plane DoFP polarization detector is used to obtain a raw intensity image based on the secondary image; it is also used to extract the observed intensity vector from the raw intensity image. ;

[0101] The imaging reconstruction module is used to reconstruct the image based on the coded pattern. The corresponding polarization modulation matrix is ​​obtained by looking up the coded polarization modulation matrix in the lookup table. ;

[0102] The coded polarization modulation matrix lookup table is obtained by any of the calibration methods for the coded polarization modulation matrix described in embodiments 1 to 4.

[0103] Also used to pass

[0104]

[0105] Obtain the Stokes vector after polarization error correction ,in, The polarization modulation matrix The pseudo-inverse matrix;

[0106] Also used based on the Stokes vector By using imaging reconstruction methods, a polarization image of the target scene with polarization error removed is obtained, thus completing the correction of the polarization error.

[0107] In this embodiment, the focal plane DoFP polarization detector extracts the intensity response value of each "superpixel" unit (composed of four polarization channel pixels arranged in a 2x2 pattern at 0, 45, 90, and 135) in the original intensity image. Construct the observation intensity vector under the current encoding. .

[0108] In this embodiment, the observation intensity vector is obtained through DoFP detection imaging. The process is as follows Figure 2 As shown, it includes the following steps:

[0109] Step E1: The polarization Stokes (S0, S1, S2) information emitted by the target scene is projected onto the encoding device through the telescope lens;

[0110] Step E2: Each n×n position on the encoding device corresponds to one superpixel (2x2) on the DoFP focal plane.

[0111] In step E3, a random encoding is applied to each micro-element of the n×n region, and projected onto one superpixel of the DoFP focal plane through an encoding device to obtain the intensity value. ;

[0112] For example, when n=4, a random encoding is applied to a micro-element in a 4×4 region and projected onto a superpixel on the focal plane of a DoFP through an encoding device.

[0113] In this embodiment, the intensity value of a superpixel under ideal conditions of the focal plane DoFP polarization detector is... With the incident Stokes vector The relationship should be:

[0114]

[0115] in, This is the sampling matrix of DoFP under ideal conditions.

[0116] However, in reality, considering the following relationship in the actual imaging process: Stokes vector After entering the imaging system, according to the coded pattern The Stokes vector is modulated by the encoding device to obtain the encoded and modulated vector. .

[0117] The modulation information is acquired by a DoFP polarization detector. A DoFP detector is essentially a polarimeter, which, due to its intensity... It is a Stokes vector The first component, the intensity signal output by any polarization system, is composed of the first row of the Mueller matrix M and the Stokes vector. The observation intensity vector obtained from the dot product and on DoFP for:

[0118]

[0119] The polarization modulation matrix This includes the polarization error degradation effect of the system. w0, w45, w90, and w135 are similar to the first row of the Mueller matrix of the corresponding polarizer, containing the changes in polarization state and the relationship between sampling. A matrix is ​​not only a mathematical projection, but it also acts as a "degeneracy operator".

[0120] In this embodiment, the encoding-polarization modulation matrix lookup table is used to obtain the polarization modulation matrix under different encoding patterns. .

[0121] The imaging reconstruction module modulates the polarization modulation matrix. Inverse the vector and compare it with the acquired observation intensity vector. The Stokes vector after polarization error correction was calculated. And then from The polarization information of the target scene unaffected by polarization errors is recovered.

[0122] To obtain the polarization modulation matrix Polarization calibration of the imaging system, Stokes vector Entering the imaging system, at this time modulated by the encoding device Given; under different encoding matrices, by measuring multiple sets of polarized light in different states, using known... and The encoding can be solved. W :

[0123]

[0124] Therefore, on each 2x2 superpixel of the DoFP polarization detector, and The mapping relationship can be described by W. This is achieved by measuring multiple sets of known polarization states. Combined with the data collected by the detector The polarization modulation matrix under this coding matrix can be solved using the least squares method or other inverse operations. By calibrating the polarization modulation matrix under different coding matrix modulation, a corresponding "coding-polarization modulation matrix" lookup table is established.

[0125] In this embodiment, during actual imaging, the corresponding system polarization modulation matrix can be retrieved according to the pre-constructed encoding, and the image obtained on the DoFP focal plane under this encoding can be processed using the polarization modulation matrix. Removing polarization errors. Stokes vector unaffected by polarization errors. The calculation formula is as follows:

[0126]

[0127] Represents finding the polarization modulation matrix The pseudo-inverse is obtained. The Stokes vector without polarization error is then derived. Subsequently, based on this, image reconstruction was completed using polarization computational imaging, resulting in a high-resolution polarization image of the target scene with polarization errors removed.

[0128] In this embodiment, the reconstruction method includes compressed sensing and super-resolution reconstruction.

[0129] The polarization imaging system described in this embodiment constructs a full-link polarization imaging transmission model from the target scene to the detector. Unlike traditional models, this embodiment treats each encoding step of the DMD digital micromirror array encoding module as a Mueller matrix operator that affects the polarization state, rather than a simple scalar multiplication.

Claims

1. A calibration method for a polarization-focusing plane imaging system, characterized in that, The polarization-splitting plane imaging system includes: The front-facing imaging lens projects the light field of the target scene onto the DMD digital micromirror array encoding module to form an optical image, obtaining the Stokes vector of the optical image. ; The DMD digital micromirror array encoding module can generate A coding pattern, used to employ any one coding pattern For the Stokes vector Encoding and modulation are performed, and the encoded and modulated optical image is projected onto the relay imaging lens. The Stokes vector of the light field is... , =1, 2, ..., ; The relay imaging lens projects the optical image onto the focal plane DoFP polarization detector. The focal plane DoFP polarization detector performs exposure imaging and obtains the polarization response vector. ; The calibration method includes the following steps: The DMD digital micromirror array encoding module sequentially generates one encoding pattern. For each of the coded patterns The target scene is set sequentially as follows: A uniform light field with polarization states is obtained. Group polarization parameters, based on the The polarization parameters are used to obtain the corresponding polarization modulation matrix. ; Will Each coded pattern and its corresponding The polarization modulation matrices are combined into an coded polarization modulation matrix lookup table, which serves as the calibration result of the polarization-focusing plane imaging system. pass Obtain the Stokes vector after polarization error correction ,in, The polarization modulation matrix The pseudo-inverse matrix; The Group polarization parameters include Stokes vector and its corresponding polarization response , ; The Stokes vector pass Calculated For the first The Stokes vector corresponding to a uniform optical field with polarization states, where, For band-by-band point product and integration in the spectral dimension.

2. The calibration method for the polarization-focusing plane imaging system according to claim 1, characterized in that, The A uniform optical field with various polarization states is generated using a high-precision polarization state generator.

3. The calibration method for the polarization-focusing plane imaging system according to claim 1, characterized in that, The The value is 16, and the polarization states include 4 circular polarization states and 12 linear polarization states.

4. The calibration method for the polarization-focusing plane imaging system according to claim 1, characterized in that, The The polarization state is 12, which includes 12 linear polarization states.

5. The calibration method for the polarization-focusing plane imaging system according to claim 1, characterized in that, The focal plane DoFP polarization detector exposure imaging includes extracting the intensity response values ​​of the four polarization channels of each superpixel to construct a polarization response vector. The polarization angles of the four polarization channels are 0°, 45°, 90° and 135°, respectively.

6. The calibration method for the polarization-focusing plane imaging system according to claim 1, characterized in that, The basis of The polarization parameters are used to obtain the corresponding polarization modulation matrix. The method is to minimize the error function using the least squares method. To achieve, among which, for Polarization response corresponding to group polarization parameters The matrix formed for Stokes vector corresponding to a uniform optical field in a group of polarization states The matrix formed.

7. The calibration method for the polarization-focusing plane imaging system according to claim 1, characterized in that, The polarization response It is aimed at the first A uniform light field with various polarization states, and the polarization response vector obtained during exposure imaging by a DoFP polarization detector in a focal plane.

8. A polarization-splitting plane imaging system, characterized in that, It includes a front imaging lens (2), a DMD digital micromirror array encoding module (3), a relay imaging lens (4), a focal plane DoFP polarization detector (5), and an imaging reconstruction module; The front imaging lens (2) is used to project the light field of the target scene (1) onto the DMD digital micromirror array encoding module to form an optical image, the Stokes vector of the optical image being... ; The DMD digital micromirror array encoding module is capable of generating A coding pattern, used to employ any one coding pattern For the Stokes vector The optical image is encoded and modulated, and then projected onto the relay imaging lens. The Stokes vector of this optical image is... , =1, 2, ..., ; The relay imaging lens projects the optical image onto the focal plane DoFP polarization detector; The focal plane DoFP polarization detector performs exposure imaging and obtains the polarization response vector. ; The imaging reconstruction module is used to reconstruct the image based on the coded pattern. The corresponding polarization modulation matrix is ​​obtained by looking up the coded polarization modulation matrix in the lookup table. ; The coded polarization modulation matrix lookup table is obtained by the calibration method of the coded polarization modulation matrix according to any one of claims 1 to 7; Also used to pass Obtain the Stokes vector after polarization error correction ,in, The polarization modulation matrix The pseudo-inverse matrix; Also used based on the Stokes vector The encoded and modulated Stokes vector is obtained through imaging reconstruction methods. Decoding and reconstruction are performed to obtain a polarized image of the target scene with polarization error removed, polarization error correction is completed, and a reconstructed image is obtained.

Images