Method for realizing large view field through complex wave field reconstruction based on Kramers-Kronig relation and light path thereof

Abstract

The invention discloses a method for realizing a large field of view through complex wave field reconstruction based on a Kramers-Kronig relation. Establishing a non-interference double-view-field intensity graph acquisition light path, generating sub-object light by using a spatial light modulator, and acquiring a corresponding double-view-field intensity graph; adaptively detecting an overlapping region of the double-view-field intensity graphs of the sub-object light, calculating an affine transformation matrix, and performing multiple processing to obtain a large-view-field intensity graph; calculating sub complex amplitude spectrums of the light large-view-field intensity diagrams of the sub-objects and splicing the sub complex amplitude spectrums to obtain a complete complex amplitude spectrum; inverse Fourier transform of the complete complex amplitude spectrum is calculated to obtain complete complex amplitude, and then a large-field-of-view intensity diagram and a large-field-of-view phase diagram can be obtained. According to the method, only intensity image splicing is needed, phase image splicing is not needed, the problem of phase splicing failure caused by aberration, background noise and other factors during phase splicing is solved, and the large-view-field phase image can be accurately calculated under the condition of low overlapping rate.

Description

Technical Field

[0001] The present invention relates to a large field of view reconstruction method in the field of digital holography, and specifically relates to a method and optical path for realizing a large field of view by complex wave field reconstruction based on the Kramers-Kronig relationship. Background Art

[0002] In digital holographic microscopy (DHM), how to achieve large field of view measurement while maintaining high resolution is a key issue in digital holographic microscopy technology. Currently, the method for realizing large field of view measurement generally adopts the method of separately stitching the intensity maps and phase maps reconstructed from multiple sub-field holograms.

[0003] However, the digital holographic microscopy system has inherent aberrations, such as tilt, defocus, and high-order aberrations, etc., resulting in low correlation between phase overlap regions, affecting phase registration, and further affecting the accuracy of feature detection and phase stitching. At the same time, the traditional phase stitching method uses the phase correlation method to detect the overlap region, and estimates the phase difference between signals by calculating the correlation between frequency components in the overlap region. When the overlap region is less than 50%, incomplete frequency components will appear, resulting in inaccurate estimated phase difference, causing errors in the registration point detection of the two images, resulting in failure to detect the overlap region, and thus unable to achieve effective stitching of phase images under low overlap rate conditions, and further limiting the size of the final stitched imaging field of view. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes a method and optical path for realizing a large field of view by complex wave field reconstruction based on the Kramers-Kronig relationship.

[0005] The method of the present invention only requires stitching of intensity maps and does not require stitching of phase maps, avoiding the problem of phase stitching failure caused by factors such as aberrations and background noise during the implementation of phase stitching, and can accurately calculate the large field of view phase map under low overlap rate conditions.

[0006] The technical solution of the present invention is as follows:

[0007] I. A non-interferometric dual-field intensity map acquisition optical path for large field of view complex wave field reconstruction based on the Kramers-Kronig relationship:

[0008] The non-interference dual-field intensity map acquisition optical path described above includes a laser, an adjustable attenuator G, a half-wave plate HWP, a collimating and beam expanding device BE, a microscope objective MO, a beam splitter BS, a spatial light modulator SLM, a sleeve lens TL, a first lens L1 and a second lens L2, a polarization beam splitter PBS, a first quarter-wave plate QWP1, a first mirror M1, a second quarter-wave plate QWP2, a second mirror M2, and an image sensor CMOS;

[0009] The beam emitted by the laser first passes through the adjustable attenuator G and the half-wave plate HWP in sequence, and is irradiated onto the sample to be measured through the collimating and beam expanding device BE to form an object beam. The object beam is incident on the beam splitter BS through the microscope objective MO and undergoes transmission. After transmission, it is incident on the spatial light modulator SLM, and the spatial light modulator SLM modulates the object light to generate four sub-object beams in sequence;

[0010] According to the generation order of the sub-object beams, each time a sub-object beam is generated, it propagates in the following manner: The sub-object beam is reflected by the spatial light modulator SLM and propagates back to the beam splitter BS, and then is reflected by the beam splitter BS and propagates through the sleeve lens TL and the first lens L1 in sequence and is incident on the polarization beam splitter PBS to undergo reflection and transmission, dividing the sub-object beam into two orthogonally polarized object beams O1 and O2 in two different fields of view. Among them, the vertically polarized object beam O1 passes through the first quarter-wave plate QWP1 and is retro-reflected by the first mirror M1 and returns to the polarization beam splitter PBS again to undergo transmission. The horizontally polarized object beam O2 passes through the second quarter-wave plate QWP2 and is retro-reflected by the second mirror M2 and returns to the polarization beam splitter PBS again to undergo reflection. The two orthogonally polarized object beams that return to the polarization beam splitter PBS to undergo transmission and reflection are then incident on the image sensor CMOS through the second lens L2.

[0011] The spatial light modulator SLM is an amplitude-type reflective spatial light modulator.

[0012] The collimating and beam expanding device BE is mainly composed of two lenses arranged at intervals.

[0013] The first lens L1 and the second lens L2 arranged along the optical path and their positional relationship constitute a 4F optical path system.

[0014] The spatial light modulator SLM modulates the object light to generate four sub-object beams in sequence, specifically:

[0015] Taking the optical axis center of the original object beam as the origin, sub-object beams in the up, down, left, and right directions are generated. The beam diameter of the sub-object beam is the same as that of the original object beam, and one side of each sub-object beam close to the origin is tangent to the origin.

[0016] The sub-object beam reflected at the polarization beam splitter PBS is the vertically polarized object beam O1, and the sub-object beam transmitted is the horizontally polarized object beam O2.

[0017] II. A method for realizing a large field of view based on the Kramers-Kronig relationship for complex wave field reconstruction:

[0018] The present invention does not require phase stitching, avoiding problems such as aberration affecting phase registration and inconsistent field of view baselines during the phase stitching process. The method includes sequentially performing dual-field intensity map acquisition, stitching the dual-field intensity maps to obtain a large field of view intensity map, and calculating the large field of view phase map from the large field of view intensity map using the Kramers-Kronig relationship. The method is divided into the following steps:

[0019] S1. Dual-field intensity map acquisition stage:

[0020] Build the non-interference dual-field intensity map acquisition optical path, use the spatial light modulator SLM in it to modulate the object beam passing through the sample to be measured, and sequentially generate four sub-object beams on four sides according to the time sequence. An intensity map is acquired for each sub-object beam, and a total of eight intensity maps are acquired for the four sub-object beams;

[0021] S2. Dual-field intensity map stitching stage:

[0022] Form the intensity maps of the two fields of view corresponding to each sub-object beam into a dual-field intensity map and use it as the stitching object for each stitching. Through a designed multi-algorithm fusion algorithm (Canny algorithm, cross-correlation algorithm, SIFT algorithm, cosine similarity analysis, weighted fusion algorithm), feature extraction, linear transformation, field of view registration, and weighted fusion are sequentially performed on the intensity maps of the two fields of view corresponding to each sub-object beam to obtain the stitching result of the intensity maps of the two fields of view of the sub-object beam and use it as the large field of view intensity map of the sub-object beam;

[0023] S3. Stage of obtaining the large field of view phase map from the large field of view intensity map:

[0024] After stitching the dual-field intensity maps corresponding to the sub-object beams in four directions, the large field of view intensity maps corresponding to the sub-object beams in four directions are obtained. Using the large field of view intensity maps of the four-direction sub-object beams as the input of the Kramers-Kronig relationship, the sub-complex amplitude corresponding to the large field of view intensity map of each sub-object beam is calculated by performing the Kramers-Kronig relationship; then, Fourier transform processing is performed on each sub-complex amplitude to obtain the sub-complex amplitude spectrum corresponding to each sub-object beam. The four sub-complex amplitude spectra are stitched according to the directions of their corresponding sub-object beams to obtain a complete complex amplitude spectrum. By performing inverse Fourier transform processing on the complete complex amplitude spectrum, a complete complex amplitude is obtained, and the large field of view intensity map and the large field of view phase map of the entire object beam are obtained based on the complete complex amplitude.

[0025] The stitching result of the intensity maps of two fields of view corresponding to the sub-object light is used as the large-field-of-view intensity map of the sub-object light. Four large-field-of-view intensity maps in total in four directions, namely the upper, lower, left, and right directions, will be obtained for the sub-object light in these four directions. The sub-complex amplitude spectrum in each direction is obtained by calculating the sub-complex amplitude corresponding to the large-field-of-view intensity map in each direction using the Kramers-Kronig relation and then performing a Fourier transform. When performing spectrum stitching, the sub-complex amplitude spectra in each direction are stitched according to the direction of the corresponding sub-object light.

[0026] The field-of-view ranges of the obtained large-field-of-view intensity map and large-field-of-view phase map exceed the size of the camera's sensor, enabling a larger range of the object under test to be observed in actual research.

[0027] The size of the camera's sensor determines the size of the acquired image. The smaller the size, the smaller the field of view observed by the naked eye. However, a larger field-of-view range can enable more information of the sample under test to be observed, which is beneficial for research. Through the proposed method, it is possible to expand the field-of-view range at the algorithm level, and thus achieve a larger observation range of the object under test. For example, if the size of the camera sensor is 500×500, then the size of the acquired image, that is, the maximum field-of-view size, is 500×500, resulting in too small an observation range for the sample. Through the proposed method, the field of view can be expanded, such as expanding the field-of-view size to 500×850. A larger field-of-view size can observe more field-of-view information and is also more conducive to research. The method of the present invention expands the field-of-view range at the algorithm level, rather than making the camera acquisition range larger at the hardware level.

[0028] The steps in the dual-field-of-view intensity map acquisition stage are specifically as follows:

[0029] Build a non-interference dual-field-of-view intensity map acquisition optical path. Use a spatial light modulator (SLM) to modulate the object light beam that carries sample information after passing through the sample under test and is magnified by a microscope. Generate sub-object lights in the four lateral directions of the upper, lower, left, and right with the object light beam as the center. Each sub-object light beam is split into two mutually orthogonal polarized object light beams through reflection and transmission by a polarization beam splitter (PBS). The two mutually orthogonal polarized object light beams corresponding to each sub-object light beam are sequentially acquired by an image sensor (CMOS) to obtain intensity maps. A total of eight intensity maps are acquired for the four sub-object light beams.

[0030] To ensure that the intensity maps of two fields of view corresponding to the same sub-object light are acquired, the camera is located at the same position. A 4f optical path system composed of a first lens (L1) and a second lens (L2) is introduced into the optical path to extend the image plane, and at the same time achieve the purpose of controlling the overlap rate of the two fields of view, so as to keep the camera position unchanged and achieve the purpose of successively acquiring the intensity maps of two fields of view corresponding to the same sub-object light beam by adjusting the 4f system.

[0031] In the above S1, adjust the angles of the first mirror M1 and the second mirror M2 so that two mutually orthogonal polarized object beams O1 and O2 are incident on the image sensor CMOS at different angles instead of being parallel; the two orthogonal polarized object beams O1 and O2 incident on the image sensor CMOS at different angles carry two different field-of-view information. When collecting the intensity maps, it is necessary to control the separate collection: when the image sensor CMOS collects the intensity map corresponding to the polarized object beam O1, it is necessary to block the reflection propagation channel of the second mirror M2; when the image sensor CMOS collects the intensity map corresponding to the polarized object beam O2, it is necessary to block the reflection propagation channel of the first mirror M1; specifically, any opaque object can be used for blocking.

[0032] Thus, the intensity maps of the two fields of view corresponding to the same sub-object beam are collected twice in sequence by timing.

[0033] In the above S1, adjust the reflection angles of the first mirror M1 and the second mirror M2 so that the incident angles of the two mutually orthogonal polarized object beams O1 and O2 on the image sensor CMOS are different, and finally make the fields of view / spots formed by the two mutually orthogonal polarized object beams O1 and O2 incident on the image sensor CMOS at different angles have a preset overlap rate.

[0034] The steps of the dual-field-of-view intensity map stitching stage are specifically as follows:

[0035] S21: Adaptively detect the overlapping regions of the intensity maps of the two fields of view collected for each sub-object beam.

[0036] S22: Use the SIFT algorithm to extract the feature points in the overlapping regions of the intensity maps of the two fields of view collected for each sub-object beam, and calculate and obtain the feature description matrix and its corresponding information matrix; then perform feature point screening processing based on cosine similarity analysis on the two feature description matrices of each sub-object beam to obtain the matching feature point pairs, and calculate the affine transformation matrix according to the screened matching feature point pairs, which can be used to screen out high-quality feature matching point pairs.

[0037] S23: The intensity maps of the two fields of view collected for each sub-object beam are sequentially subjected to linear transformation processing, field-of-view registration processing, and weighted fusion according to the affine transformation matrix to obtain the stitching result, which is used as the large-field-of-view intensity map of the intensity maps of the two fields of view collected for the corresponding sub-object beam.

[0038] The step S21 specifically comprises: using the Canny operator to perform edge detection on the intensity map of each field of view collected corresponding to each beam of sub-object light to obtain an edge binary map, then using the cross-correlation algorithm to calculate the cross-correlation coefficient matrix between the edge binary maps of the intensity maps of the two fields of view of each beam of sub-object light, using the pixel points with the highest positioning correlation in the cross-correlation coefficient matrix as the positioning coordinates of the overlapping area between the intensity maps of the two fields of view of each beam of sub-object light; generating a 0-1 positioning mask of the two fields of view according to the positioning coordinates of the overlapping area, and using the 0-1 positioning mask to segment the intensity maps of the two fields of view of each beam of sub-object light to obtain the overlapping areas of the intensity maps of the two fields of view of each beam of sub-object light.

[0039] Specifically, the pixel point with the highest positioning correlation in the mutual correlation coefficient matrix is ​​used as the positioning coordinate of the overlapping area between the intensity maps of the two fields of view of each sub-object light; two special coordinates (x1, y1) and (x2, y2) can be found in the positioning coordinates of the overlapping area of ​​each field of view, with x1 being the starting position of the positioning mask of 0-1, and x2 being the ending position of the positioning mask of 0-1.

[0040] The steps of obtaining a large field of view phase map from a large field of view intensity map are specifically as follows:

[0041] S31: using the Kramers-Kronig relationship, the large-field-of-view sub-complex amplitude corresponding to each sub-object beam is calculated based on the large-field-of-view intensity map obtained by stitching the intensity maps of the two fields of view collected corresponding to the sub-object beam. The large-field-of-view sub-complex amplitudes in four directions corresponding to the four sub-object beams are obtained.

[0042] S32: performing Fourier transform on the large field of view sub-complex amplitudes in four directions respectively to obtain large field of view sub-complex amplitude spectra corresponding to the respective directions;

[0043] S33: filtering and centering the large field of view sub-complex amplitude spectra in four directions in turn, and then splicing the processed large field of view sub-complex amplitude spectra according to their corresponding sub-object light directions (the directions to which the large field of view sub-complex amplitude spectra in each direction belong) to obtain a complete object light complex amplitude spectrum;

[0044] S34: Performing inverse Fourier transform on the complete object light complex amplitude spectrum to obtain a complete large-field-of-view object light complex amplitude, and extracting a large-field-of-view intensity map and a large-field-of-view phase map according to the complete large-field-of-view object light complex amplitude.

[0045] For example, we can assume that S is the complex amplitude, then the intensity map I = abs(S) 2 , phase diagram P = angle(S).

[0046] The present invention constructs an optical path for collecting non-interferometric double-field intensity maps, uses a spatial light modulator to modulate the object light to generate sub-object lights in four directions and collects the corresponding double-field intensity maps; uses the Canny operator and the cross-correlation algorithm to adaptively detect the overlapping regions of the double-field intensity maps corresponding to each beam of sub-object light and calculate the affine transformation matrix, and then performs feature extraction and matching, linear transformation, image registration and weighted fusion to obtain a large-field intensity map; uses the Kramers-Kronig relationship to calculate the sub-complex amplitudes of the large-field intensity maps of each beam of sub-object light to obtain a total of four sub-complex amplitudes, and then performs Fourier transform on the four sub-complex amplitudes to obtain their respective spectra and splice the spectra to obtain a complete complex amplitude spectrum; calculates the inverse Fourier transform of the complete complex amplitude spectrum to obtain the complete complex amplitude, and thus the large-field intensity map and the large-field phase map can be obtained.

[0047] The method of the present invention can achieve the purpose of obtaining a large-field intensity map and a large-field phase map only by performing intensity map splicing without performing phase map splicing, avoiding problems such as aberration affecting phase registration and misalignment, seams and deformation of the phase splicing result caused by inconsistent field baselines in the traditional phase splicing algorithm, and expanding the field of view while maintaining high resolution. Even in the case of a low overlap region, this method can accurately calculate the large-field phase map, which has important applications in biological tissue observation and micro-structure measurement.

[0048] The present invention innovatively combines the Canny operator and the cross-correlation algorithm in the intensity maps of the double fields corresponding to the four-direction sub-object lights collected by the non-interferometric double-field intensity map collection optical path to adaptively locate the overlapping regions between the intensity maps of the double fields corresponding to each beam of sub-object light, and uses the SIFT algorithm to extract the features of the overlapping regions of the intensity maps of the two fields corresponding to each beam of sub-object light, and then performs feature point screening based on cosine similarity to obtain high-quality matching feature point pairs, and calculates the affine transformation matrix according to the screened matching feature point pairs; then performs linear transformation processing, field registration processing, and weighted fusion by the affine transformation matrix to obtain the large-field intensity map of the intensity maps of the two fields corresponding to each beam of sub-object light, and then uses the Kramers-Kronig relationship to calculate the large-field phase map from the large-field intensity map, achieving the purpose of obtaining a large-field intensity map and a large-field phase map only by performing intensity map splicing without performing phase map splicing, avoiding problems such as aberration affecting phase registration and phase splicing failure caused by inconsistent field baselines in the process of realizing the large-field phase by phase splicing, and at the same time the method of the present invention is still effective in the case of a low overlap rate.

[0049] The technical solution of the present invention has at least achieved the following beneficial technical effects:

[0050] Compared with the traditional method for realizing a large field of view of phase, this method uses the Kramers-Kronig relationship to calculate the large field of view of the phase map based on the large field of view of the intensity map. Only the intensity map needs to be stitched, and there is no need to stitch the phase map, avoiding the problems in the traditional large field of view realization method where registration errors occur between two phase fields due to factors such as aberrations and high-frequency noise in the optical path, and the baselines of the two phase fields are not unified, resulting in the failure of phase stitching. In addition, this method can still accurately calculate the large field of view of the phase map under low overlap rates.

[0051] Generally speaking, the method of the present invention collects the dual-field intensity maps required by the algorithm through a non-interference dual-field intensity map acquisition optical path, then uses a multi-algorithm fusion method to stitch the dual-field intensity maps collected for each sub-object light beam to obtain the large field of view intensity map corresponding to that direction. Then, the Kramers-Kronig relationship is used to calculate the corresponding sub-complex amplitude according to the large field of view intensity map in each direction in turn. Then, the Fourier transform is performed on each sub-complex amplitude to obtain the corresponding sub-complex amplitude spectrum. After processing the sub-complex amplitude spectra, spectrum stitching is performed to obtain the complete complex amplitude spectrum. Then, the inverse Fourier transform is performed to obtain the complete complex amplitude, and thus the complete large field of view intensity map and large field of view phase map can be obtained. This method can accurately calculate the large field of view phase even in the presence of large aberrations and low overlap regions, effectively expanding the microscopic imaging field of view. Description of the Drawings

[0052] Figure 1 is a flowchart of the method provided in the embodiment of the present invention;

[0053] Figure 2 is an experimental result diagram of a specific example of the embodiment of the present invention.

[0054] Figure 3 For the embodiment of the present invention, the field of view intensity Figure 1 and the field of view intensity Figure 2 between different overlap rates of the large field of view stitching results. Detailed Embodiment

[0055] The present invention will be further described below in conjunction with embodiments and drawings.

[0056] In the present invention, as Figure 1 shown, a non-interference dual-field intensity map acquisition optical path is built. The non-interference dual-field intensity map acquisition optical path includes a laser, an adjustable attenuator G, a half-wave plate HWP, a collimating and beam-expanding device BE, a microscopic objective MO, a beam splitter BS, a spatial light modulator SLM, a sleeve lens TL, a first lens L1 and a second lens L2, a polarization beam splitter PBS, a first quarter-wave plate QWP1, a first mirror M1, a second quarter-wave plate QWP2, a second mirror M2, and an image sensor CMOS;

[0057] The laser beam first passes through the adjustable attenuator G and the half-wave plate HWP in sequence, and is irradiated onto the sample to be measured Sample through the collimating and beam expanding device BE to form an object beam, which is a band-limited object beam magnified by a microscope carrying sample information. Band-limited means that the diameter is limited. The object beam is incident on the beam splitter BS through the microscope objective MO and undergoes transmission. After transmission, it is incident on the spatial light modulator SLM, and the spatial light modulator SLM modulates the object light to generate four sub-object beams in sequence;

[0058] According to the generation order of the sub-object beams, each time a sub-object beam is generated, it propagates in the following manner: The sub-object beam is reflected by the spatial light modulator SLM and propagates back to the beam splitter BS, and then is reflected by the beam splitter BS and propagates through the sleeve lens TL and the first lens L1 in sequence and is incident on the polarization beam splitter PBS to undergo reflection and transmission, dividing the sub-object beam into two mutually orthogonal polarization object beams O1 and O2 containing different field-of-view information, namely the vertically polarized object beam O1 and the horizontally polarized object beam O2. Among them, the vertically polarized object beam O1 passes through the first quarter-wave plate QWP1 and is retro-reflected by the first mirror M1 and returns to the polarization beam splitter PBS to undergo transmission again, and the horizontally polarized object beam O2 passes through the second quarter-wave plate QWP2 and is retro-reflected by the second mirror M2 and returns to the polarization beam splitter PBS to undergo reflection again. The two mutually orthogonal polarization object beams that return to the polarization beam splitter PBS to undergo transmission and reflection are incident on the image sensor CMOS after passing through the second lens L2, that is, the vertically polarized object beam O1 that returns to the polarization beam splitter PBS to undergo transmission and the horizontally polarized object beam O2 that returns to the polarization beam splitter PBS to undergo reflection are combined together and then incident on the image sensor CMOS after passing through the second lens L2, and the intensity map is collected by the image sensor CMOS.

[0059] The spatial light modulator SLM modulates the object beam to generate four sub-object beams in sequence. Specifically: taking the center of the object light axis of the original object beam as the origin, sub-object beams in the upper, lower, left, and right directions are generated. The beam diameter of the sub-object beam is the same as that of the original object beam, and one side of each sub-object beam close to the origin is tangent to the origin, that is, the sub-object beams in the upper, lower, left, and right directions respectively correspond to the bottom, top, rightmost, and leftmost sides, each of which is tangent to the origin.

[0060] Therefore, after being modulated by the spatial light modulator SLM, the object beam generates sub-object beams in the upper, lower, left, and right directions in sequence.

[0061] Based on the non-interference dual-field intensity map acquisition optical path structure, the present invention is provided with a polarization beam splitter PBS, and two quarter-wave plates QWP are added at the position between the polarization beam splitter PBS and the first mirror M1 and at the position between the polarization beam splitter PBS and the second mirror M2, so as to prevent interference between the object beams O1 and O2 caused by light leakage of optical elements during intensity map acquisition, which has a negative impact on image acquisition.

[0062] The present invention introduces a 4f system into the non-interference dual-field intensity map acquisition optical path structure to extend the image plane, and at the same time achieves the purpose of controlling the overlap rate of the two fields of view, so as to keep the camera position unchanged and achieve the purpose of sequentially acquiring the intensity maps of the two fields of view by adjusting the 4f system.

[0063] The present invention constructs a non-interference dual-field intensity map acquisition optical path, uses a spatial light modulator to modulate the object beam that has passed through a microscope and carries sample information. Taking the optical axis center of the object beam as the origin, four sub-object beams in the up, down, left, and right directions are generated. Each sub-object beam carries the information of the object beam in the corresponding direction, and the intensity maps of the two fields of view are respectively acquired for each sub-object beam; the Canny operator and the cross-correlation algorithm are used to adaptively detect the overlapping regions of the intensity maps of the two fields of view acquired corresponding to each sub-object beam, the SIFT algorithm is used to extract the features of the overlapping regions of the intensity maps of the two fields of view acquired corresponding to each sub-object beam, and then high-quality matching feature point pairs are obtained through feature point screening based on cosine similarity, and the affine transformation matrix is calculated according to the obtained matching feature point pairs; the intensity maps of the two fields of view acquired corresponding to each sub-object beam are sequentially subjected to linear transformation processing, field-of-view registration processing, and weighted fusion according to the affine transformation matrix to obtain the large-field intensity map of the intensity maps of the two fields of view acquired corresponding to the sub-object beam; then the Kramers-Kronig relationship is used to calculate the large-field phase map from the large-field intensity map.

[0064] The present invention innovatively combines the Canny operator and the cross-correlation algorithm to adaptively locate the overlapping area between the intensity maps of the dual fields of view collected by each beam of sub-object light in the intensity maps of the dual fields of view collected by the sub-object light in four directions recorded by the non-interference dual-field-of-view intensity map collection optical path, and uses the SIFT algorithm to extract the features of the overlapping area of ​​the intensity maps of the two fields of view collected by each beam of sub-object light, and then performs feature point screening based on cosine similarity to obtain high-quality matching feature point pairs, and calculates the affine transformation matrix based on the screened matching feature point pairs; and then performs linear transformation based on the affine transformation matrix. The large field of view intensity map of the intensity maps of the two fields of view collected corresponding to each beam of sub-object light is obtained by performing property transformation processing, field of view registration processing, and weighted fusion, and then the large field of view phase map is calculated from the large field of view intensity map by using the Kramers-Kronig relationship, so that only intensity map splicing is required, and phase map splicing is not required to be performed to obtain the large field of view intensity map and the large field of view phase map, and the problems of aberration affecting phase registration and phase splicing failure caused by inconsistent field of view baselines in the process of phase splicing to achieve large field of view phase are avoided. At the same time, the method of the present invention is still effective under low overlap rate conditions.

[0065] Embodiments of the present invention Figure 1 As shown in the flowchart, the specific process is as follows:

[0066] (1) Dual-field intensity map acquisition stage

[0067] A non-interference dual-field intensity map collection optical path is built, and the spatial light modulator SLM therein is used to modulate the band-limited objective light beam that carries the sample information of the sample to be tested and is amplified by a microscope. Four sub-object light beams are generated on four sides in sequence, and an intensity map is obtained by collecting each sub-object light beam. A total of eight intensity maps are collected by the four sub-object light beams.

[0068] In the non-interference dual-field intensity map collection optical path, the light beam emitted by the He-Ne laser first passes through the adjustable attenuator G and the half-wave plate HWP in sequence, irradiates the measurement sample Sample through the collimating beam expander BE, and then enters the microscope objective lens MO to form a band-limited object light, and then passes through the beam splitter BS to enter the spatial light modulator SLM. The spatial light modulator SLM modulates the band-limited object light, and generates sub-object lights in four directions of up, down, left, and right with the optical axis center of the object light beam as the origin. Each time a beam of sub-object light is generated, the sub-object light is reflected by the spatial light modulator SLM and propagates to the beam splitter BS, and then enters the sleeve lens after being reflected by the beam splitter BS, and then enters the 4f system composed of the first lens L1 and the second lens L2.

[0069] A polarization beam splitter PBS is set in the 4f system. The polarization beam splitter PBS divides the sub-object light into two mutually orthogonal polarized object beams O1 and O2 containing different field-of-view information. Among them, the vertically polarized object beam O1 passes through the first quarter-wave plate QWP1 and is reflected back to the polarization beam splitter PBS by the first mirror M1 and then transmitted. The horizontally polarized object beam O2 passes through the second quarter-wave plate QWP2 and is reflected back to the polarization beam splitter PBS by the second mirror M2 and then reflected. The orthogonal polarized object beams O1 and O2 that are transmitted and reflected back to the polarization beam splitter PBS are then incident on the image sensor CMOS through the second lens L2 to obtain a dual-field intensity map.

[0070] When collecting the dual-field intensity map, the overlapping area size of the two fields of view is controlled by adjusting the reflection angles of the first mirror M1 and the second mirror M2. At the same time, when collecting the intensity map corresponding to the vertically polarized object beam O1, the propagation path of the horizontally polarized object beam O2 in the 4f system is blocked; when collecting the intensity map corresponding to the horizontally polarized object beam O2, the propagation path of the vertically polarized object beam O1 in the 4f system is blocked. In this way, the intensity maps corresponding to the orthogonal polarized object beams O1 and O2 can be obtained respectively.

[0071] Specifically, the spatial light modulator SLM is used to modulate the object beam carrying sample information and magnified by the microscope after passing through the sample to be measured. Taking the optical axis center of the object beam as the origin, sub-object lights in four directions, namely up, down, left, and right, are generated. Each sub-object light is reflected and transmitted by the polarization beam splitter PBS to be divided into two mutually orthogonal polarized object beams. The two mutually orthogonal polarized object beams corresponding to each sub-object light are sequentially collected by the image sensor CMOS for the field of view to obtain the intensity map. A total of eight intensity maps are collected for the four sub-object lights.

[0072] Adjust the angles of the first mirror M1 and the second mirror M2 so that the two mutually orthogonal polarized object beams O1 and O2 are incident on the image sensor CMOS at different angles, rather than parallel incident.

[0073] The two orthogonal polarized object beams O1 and O2 incident on the image sensor CMOS at different angles carry two different field-of-view information. When collecting the intensity map, it is necessary to control the separate collection:

[0074] When the image sensor CMOS collects the intensity map corresponding to the polarized object beam O1, it is necessary to block the reflection propagation channel of the second mirror M2.

[0075] When the image sensor CMOS collects the intensity map corresponding to the polarized object beam O2, it is necessary to block the reflection propagation channel of the first mirror M1.

[0076] In this way, the intensity maps of the two fields of view corresponding to the same sub-object light are collected twice in sequence.

[0077] Figure 2 (a) is a schematic diagram of the spatial light modulator SLM modulating the band-limited object light to generate the sub-object light and the corresponding modulation aperture. The solid circle represents the band-limited object light, and the four dashed circles above, below, left, and right represent the sub-object lights in four directions generated by modulating the spatial light modulator SLM with the optical axis center of the object light beam as the origin; Figure 2 (b, c, d, e) and Figure 2 (b1, c1, d1, e1) are the intensity maps corresponding to the sub-object lights in four directions collected in two fields of view respectively; among them Figure 2 (b, b1), Figure 2 (c, c1), Figure 2 (d, d1), Figure 2 (e, e1) are the intensity maps of two fields of view corresponding to the same beam of sub-object light in the four directions of up, down, left, and right respectively.

[0078] (2) Stage of stitching the double-field-of-view intensity maps to obtain the large-field-of-view intensity map

[0079] First, use the Canny operator to perform edge detection on the intensity map of each field of view corresponding to each beam of sub-object light to obtain the edge binary map, and then use the cross-correlation algorithm to calculate the cross-correlation coefficient matrix between the edge binary maps of the intensity maps of two fields of view of each beam of sub-object light. Use the pixel point with the highest positioning correlation degree in the cross-correlation coefficient matrix as the positioning coordinates of the overlapping area between the intensity maps of two fields of view of each beam of sub-object light.

[0080] Generate the 0-1 positioning mask of the two fields of view according to the positioning coordinates of the overlapping area, and use the 0-1 positioning mask to segment the intensity maps of two fields of view of each beam of sub-object light to obtain the overlapping areas of the intensity maps of two fields of view of each beam of sub-object light respectively.

[0081] The cross-correlation coefficient matrix R(u, v) is expressed as:

[0082]

[0083] In the formula, M×N is the image size; I m1 and I m2 are the edge binary maps of field of view 1 and field of view 2 respectively; (x i , y i ) is the pixel coordinate in I m1 ; (u, v) is the displacement amount between (x i , y i ) and the center of the sliding window, 0 ≤ u ≤ M - 1, 0 ≤ v ≤ N - 1; and are respectively centered on (x i , y i ) and (x i + u, yi The average pixel intensity within a sliding window centered on (+v); the size of the cross-correlation coefficient matrix R is (M×N). 2 .

[0084] Then, the SIFT algorithm is used to extract the features of the overlapping region of the intensity maps of the two fields of view corresponding to each sub-object beam, and then feature point screening based on cosine similarity is performed to obtain high-quality matching feature point pairs. In the cosine similarity matrix, only elements greater than the empirical threshold are retained and indexed in the information matrix to obtain the coordinate sets (X1, Y1) and (X2, Y2) of the feature matching points between the intensity map field of view 1 and the intensity map field of view 2, and the affine transformation matrix is calculated based on the screened matching feature point pairs.

[0085] The SIFT algorithm extracts the feature points in the overlapping region of the intensity maps of field of view 1 and field of view 2, and calculates the feature description matrix and the corresponding information matrix of each field of view intensity map. Since the feature points in the overlapping region of field of view 1 and field of view 2 are not necessarily in one-to-one correspondence, further cosine similarity analysis is performed to screen out the feature point pairs with high feature point matching degrees in the overlapping region of the two fields of view.

[0086] Least squares calculation is performed on the matching feature point pairs to obtain the affine transformation matrix, expressed as:

[0087]

[0088] In the formula, (a, b, c, d, e, f) are the parameters of the affine transformation matrix to be solved.

[0089] Finally, through linear transformation, the intensity map field of view 1 and the intensity map field of view 2 are field registered using the calculated affine transformation matrix to achieve image stitching of the two intensity map fields of view, and then the weighted fusion algorithm is used to process the stitched image to achieve seamless smooth transition.

[0090] Figure 2 (b2, c2, d2, e2) are respectively Figure 2 (b, b1), Figure 2 (c, c1), Figure 2 (d, d1), Figure 2 (e, e1) The large field of view intensity maps stitched from the intensity maps of the two fields of view corresponding to each sub-object beam collected in the four sub-apertures.

[0091] (3) The stage of calculating the large field of view phase map from the large field of view intensity map using the Kramers-Kronig relationship

[0092] Since the optical re - wave field is a complex number, and the object - light information after passing through the microscope objective is a band - limited signal, satisfying the analyticity in the frequency domain. The light passing through the sample can be decomposed into the superposition of the unscattered plane wave and the sample - scattered wave. Therefore, the intensity map I(x, y) can be expressed as the superposition of these two:

[0093] I(x,y)=|F -1 {S(u,v)}| 2 =|F -1 {S'(u,v)+δ(u,v)}| 2 =|s'(x,y)+e -j(0x+0y) | 2

[0094] Where, F -1 {·} represents the inverse Fourier transform, S(u, v) represents the complex - amplitude spectrum, S'(u, v) represents the spectrum of the scattered wave exiting from the sample plane, δ(u, v) represents the spectrum of the unscattered plane wave passing through the sample, s'(x, y) represents the scattered wave exiting from the sample plane, e -j(0·x+0·y ) represents the unscattered plane wave passing through the sample, j represents the imaginary - function flag, (u, v) represents the frequency - domain coordinates, and (x, y) represents the time - domain coordinates.

[0095] Since the Fourier transform has the frequency - shift property, when the spectrum S(u, v) located at the center of the frequency - domain coordinates is frequency - shifted, only the phase term will be multiplied by the original spectrum. However, the camera cannot detect the phase change and can still measure the same intensity image. Through Figure 1 Step 2 in the implementation flowchart of the method of the present invention shown, four large - field - of - view intensity maps I i (x,y), (i = 1,2,3,4) corresponding to the four - direction sub - object lights can be obtained. Here, taking I1(x,y) as an example, then I1(x,y) can be expressed as:

[0096] I1(x,y)=|F -1 {S1(u,v)}| 2 =|F -1 {S1(u - u1,v - v1)}| 2

[0097] Where, S1(u - u1,v - v1) represents the complex - amplitude spectrum obtained after S1(u, v) is frequency - shifted by (u1, v1), and F -1 {·} represents the inverse Fourier transform.

[0098] Then S1(u - u1,v - v1) can be separated into the sum of the scattered wave and the unscattered wave, that is:

[0099] S1(u - u1, v - v1) = S1'(u - u1, v - v1) + δ1(u - u1, v - v1)

[0100] where S1'(u - u1, v - v1) represents the scattered wave spectrum obtained by frequency - shifting the scattered wave spectrum S1'(u, v) emitted from the sample plane, and δ1(u - u1, v - v1) represents the unscattered plane - wave spectrum obtained by frequency - shifting the unscattered plane - wave spectrum δ1(u, v) passing through the sample. Let S 11 (u, v) = S1'(u - u1, v - v1), then I1(x, y) can be expressed as:

[0101] I1(x, y) = |F -1 {S 11 (u, v)+δ1(u - u1, v - v1)}| 2

[0102] Generate a hypothetical reference plane - wave r1(x, y):

[0103]

[0104] where r1(x, y) represents the reference plane - wave generated from the frequency - shifted complex - amplitude spectrum S1(u - u1, v - v1) obtained by frequency - shifting the complex - amplitude spectrum S1(u, v) corresponding to I1(x, y) during the process of calculating the large - field phase map from the large - field intensity map I1(x, y) using the Kramers - Kronig relation. j represents the imaginary - function flag, and e {·} represents the exponential with base e.

[0105] Then specify a Hilbert kernel:

[0106] H1(u, v) = - jsgn(v1)·sgn(v)

[0107] where H1(u, v) represents the Hilbert kernel generated from the frequency - shifted complex - amplitude spectrum S1(u - u1, v - v1) obtained by frequency - shifting the complex - amplitude spectrum S1(u, v) corresponding to I1(x, y) during the process of calculating the large - field phase map from the large - field intensity map I1(x, y) using the Kramers - Kronig relation. v1 represents the frequency - shift distance on the v - axis in the frequency domain of the complex - amplitude spectrum S1(u, v), and sgn(·) represents the sign function. So:

[0108] I1(x, y) = |s 11 (x, y)+r1(x, y)| 2

[0109] Assume an intermediate variable X1:

[0110]

[0111] Among them, ln{·} represents taking the logarithm, Re{X1} represents the real part of X1, Im{X1} represents the imaginary part of X1, and i represents the imaginary function flag. In actual measurement, the real part Re{X1} of X1 can be expressed as an intensity measurement, which refers to the large-field intensity map in the method of the present invention. The imaginary part Im{X1} of X1 can be calculated from the large-field intensity map by using the Kramers-Kronig relation. Therefore, the imaginary part of X1 can be expressed as:

[0112] Im{X1} = F -1 {F{Re{X1}}·H1(u, v)}

[0113] Among them, F{·} represents the Fourier transform. Therefore, the complex amplitude spectrum can be expressed as:

[0114]

[0115] Among them, Re{X1} represents the real part of X1, Im{X1} represents the imaginary part of X1, r1(x, y) represents a reference plane wave generated from the shifted complex amplitude spectrum S1(u - u1, v - v1) obtained after frequency shifting (u1, v1) of the complex amplitude spectrum S1(u, v) corresponding to I1(x, y) during the process of calculating the large-field phase map of I1(x, y) from the large-field intensity map I1(x, y) using the Kramers-Kronig relation, j represents the imaginary function flag, and e {·} represents the exponential with base e.

[0116] Shifting S 11 (u - u1, v - v1) can obtain the complex amplitude spectrum S1(u, v) corresponding to the large-field intensity map I1(x, y). After the four large-field intensity maps are processed by the Kramers-Kronig relation calculation, four sub-complex amplitude spectra S1(u, v), S2(u, v), S3(u, v), and S4(u, v) are obtained. The four sub-complex amplitude spectra are spectrally stitched according to their respective corresponding sub-object light directions to obtain a complete complex amplitude spectrum, and then the inverse Fourier transform of the complete complex amplitude spectrum is calculated to obtain the complete complex amplitude. Based on the complete complex amplitude, the large-field intensity map and the large-field phase map can be further obtained.

[0117] Figure 2 (b3, c3, d3, e3) are respectively the four corresponding sub-complex amplitude spectra calculated from the large-field intensity maps shown by Figure 2 (b2, c2, d2, e2) through the Kramers-Kronig relation, Figure 2(f) is the complete complex amplitude spectrum obtained by splicing the four sub-complex amplitude spectra in the corresponding directions. Figure 2 (g) is Figure 2 (f) the complete complex amplitude obtained by performing the inverse Fourier transform. Figure 2 (h) and Figure 2 (i) are respectively the large field-of-view intensity map and the large field-of-view phase map calculated according to the complete complex amplitude.

[0118] As can be seen from the implementation, the present invention utilizes the characteristic that the Kramers-Kronig relationship can describe the connection between the real part and the imaginary part of a complex function. By collecting the double field-of-view intensity maps through a non-interference optical path, then using the Canny operator combined with the cross-correlation algorithm to adaptively detect the double field-of-view intensity maps collected corresponding to each beam of sub-object light and locate the overlapping region, and performing feature extraction on the overlapping region to calculate the affine transformation matrix, and then obtaining the large field-of-view intensity map through linear transformation, field-of-view registration, and weighted fusion. Then, using the Kramers-Kronig relationship, the corresponding sub-complex amplitude spectrum is calculated according to the large field-of-view intensity map corresponding to each beam of sub-object light, and then the spectra are spliced to obtain the complete complex amplitude spectrum. By performing the inverse Fourier transform on the complete complex amplitude spectrum, the complete complex amplitude is obtained, and the large field-of-view intensity map and the large field-of-view phase map are calculated according to the complete complex amplitude, achieving the purpose of obtaining the large field-of-view intensity map and the large field-of-view phase map only by intensity map splicing without phase map splicing, avoiding the situation of phase splicing failure caused by reasons such as aberration, coherent noise, limited overlapping region, low correlation of the phase maps to be spliced, and lack of a common baseline in the traditional method of realizing a large phase field.

[0119] Using the method of the present invention, even in the presence of aberration, coherent noise, and low overlap rate, the large field-of-view intensity map and the large field-of-view phase map can be accurately calculated without obvious splicing gaps, with good splicing accuracy and stability, effectively expanding the field of view of digital holographic microscopy.

[0120] The method of the present invention only needs to perform intensity map splicing without performing phase map splicing, avoiding problems such as phase registration caused by aberration and inconsistent field-of-view baselines during phase splicing, and can accurately calculate the large field-of-view phase map even in the case of low overlap rate. Figure 3Shown are the results obtained by verifying the effectiveness of the method for different overlap rates when the field-of-view sizes of the field-of-view 1 intensity map and the field-of-view 2 intensity map are 1000×1000 and 500×500 respectively using this method. It can be concluded that by using this method, successful stitching can be achieved even when the overlap rate is 11%, and the large-field phase map can be calculated from the successfully stitched large-field intensity map using the Kramers-Kronig relation. However, when the overlap rate is lower than 11%, the overlapping area of the two fields of view is too small, resulting in the Canny operator being unable to accurately obtain the overlapping region of the two fields of view, and further causing the SIFT algorithm to be unable to calculate the feature points in the overlapping region of the two fields of view for the affine matrix to complete the accurate mapping of the feature points. Therefore, the field-of-view stitching fails at this time. The method of the present invention can achieve a maximum field-of-view expansion of 1.89 times.

[0121] The above specific embodiments are used to explain and illustrate the present invention, rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims of the present invention fall within the protection scope of the present invention. The above are only the preferred embodiments of the present invention. Therefore, all equivalent changes or modifications made according to the structure, features, and principles described in the scope of the present invention patent application are included in the scope of the present invention patent application.

Claims

1. A complex wave field reconstruction large field of view non-interference dual field of view intensity image acquisition optical path based on Kramers-Kronig relationship, characterized by: The non-interference dual-field intensity map collection optical path includes a laser, an adjustable attenuator G, a half-wave plate HWP, a collimating beam expander BE, a microscope objective lens MO, a beam splitter BS, a spatial light modulator SLM, a sleeve lens TL, a first lens L1 and a second lens L2, a polarization beam splitter PBS, a first quarter-wave plate QWP1, a first reflector M1, a second quarter-wave plate QWP2, a second reflector M2 and an image sensor CMOS; the light beam emitted by the laser first passes through the adjustable attenuator G and the half-wave plate HWP in sequence, and is irradiated to the sample to be tested through the collimating beam expander BE to form an object beam, and the object beam is incident on the beam splitter BS through the microscope objective lens MO for transmission, and then is incident on the spatial light modulator SLM after transmission, and the spatial light modulator SLM modulates the object light and generates four sub-object beams in sequence; According to the generation order of sub-object light, each generated sub-object light is propagated in the following manner: the sub-object light is reflected by the spatial light modulator SLM and propagated back to the beam splitter BS, and then after being reflected by the beam splitter BS, it is successively propagated through the sleeve lens TL and the first lens L1 and then incident on the polarization beam splitter PBS for reflection and transmission, and the sub-object light is divided into two orthogonal polarized object light beams O1 and O2 with two different fields of view, wherein the vertically polarized object light beam O1 passes through the first quarter wave plate QWP1, is reversely reflected by the first reflector M1, and then returns to the polarization beam splitter PBS for transmission again, and the horizontally polarized object light beam O2 passes through the second quarter wave plate QWP2, is reversely reflected by the second reflector M2, and then returns to the polarization beam splitter PBS for reflection again, and the two orthogonal polarized object light beams that return to the polarization beam splitter PBS for transmission and reflection again pass through the second lens L2 and are incident on the image sensor CMOS.

2. According to claim 1, a complex wave field reconstruction large field non-interference dual field intensity map acquisition optical path based on Kramers-Kronig relationship, characterized in that: The spatial light modulator SLM modulates the object light to generate four sub-object light beams in sequence. Specifically, with the optical axis center of the original object light beam as the origin, sub-object light beams in four directions of up, down, left and right are generated. The beam diameter of the sub-object light beam is the same as the beam diameter of the original object light beam, and the side of each sub-object light beam close to the origin is tangent to the origin.

3. According to claim 1, a non-interference dual-field intensity image acquisition optical path for complex wave field reconstruction of a large field of view based on the Kramers-Kronig relationship, characterized in that: The sub-object beam reflected at the polarization beam splitter PBS is a vertically polarized object beam O1, and the sub-object beam transmitted is a horizontally polarized object beam O2.

4. A method for realizing a large field of view by complex wavefield reconstruction based on the Kramers-Kronig relationship, applied to the non-interference dual-field-of-view intensity map acquisition optical path of claim 1, characterized in that: The method is divided into the following steps: S1. Dual-field intensity map acquisition stage: The non-interference dual-field intensity map collection optical path is constructed, and the spatial light modulator SLM therein is used to modulate the object light beam passing through the sample to be tested, and four sub-object light beams on four sides are generated in sequence, and each sub-object light beam is collected to obtain an intensity map, and a total of eight intensity maps are collected by the four sub-object light beams; S2, dual field intensity map stitching stage: The two intensity images corresponding to each sub-object light beam are respectively used to form a dual-field intensity image and used as the stitching object for each stitching. The two intensity images corresponding to each sub-object light beam are sequentially subjected to feature extraction, linear transformation, field of view registration, and weighted fusion through a multi-algorithm fusion algorithm to obtain the stitching result of the two intensity images of the sub-object light and use it as the large-field intensity image of the sub-object light. S3, the stage of obtaining the phase image of the large field of view from the intensity image of the large field of view: The large-field intensity map of four sub-object beams is used as the input of the Kramers-Kronig relationship, and the sub-complex amplitude corresponding to the large-field intensity map of each sub-object beam is calculated by executing the Kramers-Kronig relationship; each sub-complex amplitude is then Fourier transformed to obtain the sub-complex amplitude spectrum corresponding to each sub-object beam, and the four sub-complex amplitude spectra are spectrally spliced ​​according to the corresponding sub-object beam directions to obtain a complete complex amplitude spectrum, and the complete complex amplitude is obtained by calculating the inverse Fourier transform of the complete complex amplitude spectrum, and the large-field intensity map and large-field phase map of the entire object light are obtained according to the complete complex amplitude.

5. The method for realizing a large field of view by complex wavefield reconstruction based on Kramers-Kronig relationship according to claim 4, characterized in that: The steps of the dual-field intensity map acquisition stage are specifically as follows: A non-interference dual-field intensity map collection optical path is built, and the spatial light modulator SLM is used to modulate the object beam that carries sample information after passing through the sample to be tested and is magnified by a microscope. Sub-object beams are generated in the four directions of up, down, left, and right with the object beam as the center. Each sub-object beam is reflected and transmitted by a polarization beam splitter PBS to be divided into two mutually orthogonal polarized object beams. The two mutually orthogonal polarized object beams corresponding to each sub-object beam are collected in turn by the image sensor CMOS to obtain the intensity map of the field of view. A total of eight intensity maps are collected by the four sub-object beams.

6. The method for realizing a large field of view by complex wavefield reconstruction based on Kramers-Kronig relationship according to claim 4, characterized in that: In S1, the angles of the first reflector M1 and the second reflector M2 are adjusted so that the two orthogonal polarized light beams O1 and O2 are incident on the image sensor CMOS at different angles; The two orthogonal polarized object beams O1 and O2 incident on the image sensor CMOS at different angles carry two different field of view information, which are controlled to be collected separately when the intensity map is collected: When collecting the intensity graph corresponding to the polarized object light beam O1, the reflection propagation channel of the second reflector M2 is blocked; when collecting the intensity graph corresponding to the polarized object light beam O2, the reflection propagation channel of the first reflector M1 is blocked; the two intensity graphs corresponding to the same beam of object light are collected twice in sequence.

7. The method for realizing a large field of view by complex wavefield reconstruction based on Kramers-Kronig relationship according to claim 4, characterized in that: In S1, the reflection angles of the first reflector M1 and the second reflector M2 are adjusted so that the two orthogonal polarized light beams O1 and O2 are incident on the image sensor CMOS at different angles, and finally the field of view / light spot formed by the two orthogonal polarized light beams O1 and O2 incident on the image sensor CMOS at different angles has a preset overlap rate.

8. The method for realizing a large field of view by complex wavefield reconstruction based on Kramers-Kronig relationship according to claim 4, characterized in that: The steps of the dual-field intensity map stitching stage are specifically as follows: S21: adaptively detecting the overlapping area of ​​two intensity images collected corresponding to each sub-object light beam; S22: extracting feature points of the overlapping area of ​​the two intensity images collected corresponding to each sub-object light beam by using the SIFT algorithm, and calculating a feature description matrix and its corresponding information matrix; then performing feature point screening based on cosine similarity analysis on the two feature description matrices of each sub-object light beam to obtain matching feature point pairs, and calculating an affine transformation matrix based on the screened matching feature point pairs; S23: The two intensity images collected corresponding to each sub-object light beam are sequentially subjected to linear transformation processing, field of view registration processing, and weighted fusion processing according to the affine transformation matrix to obtain a stitching result as a large field of view intensity image of the two intensity images collected corresponding to the sub-object light beam.

9. The method for realizing a large field of view by complex wavefield reconstruction based on Kramers-Kronig relationship according to claim 4, characterized in that: The step S21 is specifically as follows: The Canny operator is used to perform edge detection on each intensity image collected corresponding to each sub-object light beam to obtain an edge binary image. Then, the cross-correlation algorithm is used to calculate the cross-correlation coefficient matrix between the edge binary images of the two intensity images of each sub-object light beam. The pixel point with the highest positioning correlation in the cross-correlation coefficient matrix is ​​used as the positioning coordinate of the overlapping area between the two intensity images of each sub-object light beam. A 0-1 positioning mask of the two fields of view is generated according to the positioning coordinates of the overlapping area, and the two intensity maps of each sub-object light beam are segmented by the 0-1 positioning mask to obtain the overlapping area of ​​the two intensity maps of each sub-object light beam.

10. The method for realizing a large field of view by complex wavefield reconstruction based on Kramers-Kronig relationship according to claim 1, characterized in that: The steps of obtaining a large field of view phase map from a large field of view intensity map are specifically as follows: S31: Using the Kramers-Kronig relationship, the corresponding large-field sub-complex amplitude is calculated according to the large-field intensity diagram corresponding to each sub-object beam, and the large-field sub-complex amplitudes in four directions corresponding to the four sub-object beams are obtained; S32: performing Fourier transform on the large field of view sub-complex amplitudes in four directions respectively to obtain large field of view sub-complex amplitude spectra corresponding to the respective directions; S33: filtering and centering the large field of view sub-complex amplitude spectra in four directions in turn, and then splicing the processed large field of view sub-complex amplitude spectra according to their corresponding sub-object light directions to obtain a complete object light complex amplitude spectrum; S34: Performing inverse Fourier transform on the complete object light complex amplitude spectrum to obtain a complete large-field-of-view object light complex amplitude, and extracting a large-field-of-view intensity map and a large-field-of-view phase map according to the complete large-field-of-view object light complex amplitude.

Images