A wide-spectrum partially-coherent diffractive computational imaging method and system

Abstract

This invention provides a broadband partially coherent diffraction computational imaging method and system, belonging to the field of coherent diffraction computational imaging technology. The method includes: incident a broadband beam onto the sample to be tested; establishing a broadband diffraction image monochromatic matrix equation based on the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam and the broadband partially coherent diffraction intensity distribution in the far field of Fraunhofer diffraction under the sample to be tested; and calculating the monochromatic diffraction intensity distribution; substituting the monochromatic diffraction intensity distribution into the coherent diffraction computational imaging algorithm to output the reconstructed result of the coherent diffraction computational imaging. This invention suppresses the decoherence phenomenon caused by broadband diffraction, improving the quality and convergence robustness of broadband partially coherent diffraction computational imaging.

Description

Technical Field

[0001] This invention belongs to the field of coherent diffraction computational imaging technology, and more specifically, relates to a broadband partially coherent diffraction computational imaging method and system. Background Technology

[0002] Coherent diffraction imaging is a relatively advanced lensless computational imaging technique in the fields of biological imaging and micro / nano measurement. Existing coherent diffraction imaging systems mainly employ a coherent light source that converges onto the sample surface to form a micro-spot. After transmission or reflection from the sample, coherent diffraction occurs, propagating to the far field where the diffracted light intensity is captured by a detector. The complex amplitude function of the sample is then reconstructed using a phase retrieval computational imaging algorithm. To meet the high coherence requirement of the illumination source, monochromatic spectral modulation of the incident light is usually performed, such as by adding a bandpass filter, which significantly reduces the photon utilization efficiency of the light source. Using a broadband light source can enhance the photon utilization efficiency across the entire spectral range; however, spectral broadening naturally introduces decoherence due to diffraction aliasing, thus hindering the successful convergence of coherent diffraction imaging.

[0003] In recent years, broadband partially coherent diffraction computational imaging techniques have seen the emergence of novel algorithms, such as mixed-state multi-wavelength spectral decomposition. This method discretizes the input spectrum and then uses a complex algorithm to traverse the entire discrete spectrum, iteratively calculating the energy weights of each spectral component. These solutions face several insurmountable challenges, including complex iterative calculations, the need to traverse dense wavelength channels throughout the complete spectrum, the requirement for accurate spectral measurements as input, strict constraints on the non-dispersive sample across the entire spectral range, and the requirement for the solution to converge within bandwidth limitations to be effective. These problems severely hinder the progress of broadband partially coherent diffraction computational imaging techniques. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a broadband partially coherent diffraction computational imaging method and system, which addresses the problems of high requirements for light source coherence in existing coherent diffraction imaging technologies, limited spectral bandwidth in broadband partially coherent diffraction methods, and high algorithm complexity.

[0005] To achieve the above objectives, the present invention provides a broadband partially coherent diffraction computational imaging method, comprising the following steps:

[0006] A broadband beam is incident on the sample to be tested. Based on the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam and the coherent diffraction intensity distribution of the broadband part of the Fraunhofer diffraction far field under the sample to be tested, the monochromatic diffraction image monochromatic matrix equation is established, and the monochromatic diffraction intensity distribution is obtained.

[0007] Substitute the monochromatic diffraction intensity distribution into the coherent diffraction computational imaging algorithm to output the reconstruction result of coherent diffraction computational imaging;

[0008] The method for obtaining the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam is as follows:

[0009] A broadband beam is incident on a reference sample through an incident optical path to obtain the coherent diffraction intensity distribution of the broadband far-field Fraunhofer diffraction under the reference sample; wherein, the optical materials of the reference sample and the sample to be tested are the same;

[0010] A bandpass filter is used to monochromaticly filter and modulate a broadband beam into a monochromatic beam. The monochromatic beam is then incident on a reference sample through the incident optical path to obtain the coherent diffraction intensity distribution of the far-field monochromatic beam of Fraunhofer diffraction under the reference sample.

[0011] By combining the intensity distribution of partially coherent far-field wide-spectrum Fraunhofer diffraction and the intensity distribution of coherent diffraction of monochromatic beam under the reference sample, a multivariate linear regression model of wide-spectrum diffraction is established to solve the discrete spectral energy density.

[0012] The diffusion and transfer matrix of the diffraction spectral points corresponding to the monochromatic beam is established based on the discrete spectral energy density.

[0013] More preferably, the incident form of the incident light path is: using an aperture device to cut the broadband light beam into a light spot and incident it onto the reference sample or the sample to be tested;

[0014] Alternatively, a focusing lens can be used to converge a broadband beam of light into a spot and direct it onto the reference sample or the sample to be tested.

[0015] More preferably, the broadband diffraction multiple linear regression model is as follows:

[0016] B0=ω T ·X λ +b

[0017]

[0018] in, For the reference sample in the wavelength matrix λ=[λ0,λ1,λ2,…,λ n The diffraction intensity distribution matrix under monochromatic coherent beam incident conditions; ω represents the spectral weight corresponding to the discrete wavelength λ; b is the system noise error component; AT is the linear transformation transfer function; and the scaling transformation coefficient is... λ c The wavelength of a monochromatic beam; m c B0 represents the coherent diffraction intensity distribution of the monochromatic beam in the far field of Fraunhofer diffraction; B0 represents the coherent diffraction intensity distribution of the broadband far field of Fraunhofer diffraction; λ0, λ1, λ2, ..., λ n The wavelengths of the broadband beam are discrete in the spectrum.

[0019] More preferably, the monochromatic matrix equation for the broadband diffraction image is C T B = C T Cm; where C is the diffusion transfer matrix of the diffraction spectral point corresponding to the monochromatic beam; B is the intensity distribution of the broad-spectral coherent diffraction light under the sample; and m is the intensity distribution of the monochromatic diffraction light.

[0020] More preferably, the coherent diffraction computational imaging algorithm includes a coherent diffraction imaging algorithm, a stacked diffraction imaging algorithm, a Fourier stacked algorithm, and a computational holographic imaging algorithm.

[0021] More preferably, the spectral bandwidth of the broadband beam ranges from the extreme ultraviolet band to the far-infrared band.

[0022] On the other hand, the present invention provides a broadband partially coherent diffraction computational imaging system, comprising:

[0023] The monochromatic diffraction intensity distribution calculation module is used to incident a broadband beam onto the sample to be tested. Based on the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam and the coherent diffraction intensity distribution of the broadband part of the Fraunhofer diffraction far field under the sample to be tested, the monochromatic diffraction image monochromatic matrix equation is established, and the monochromatic diffraction intensity distribution is obtained.

[0024] The reconstruction module of coherent diffraction computational imaging is used to substitute the monochromatic diffraction intensity distribution into the coherent diffraction computational imaging algorithm and output the reconstruction result of coherent diffraction computational imaging.

[0025] The module for acquiring broadband partially coherent diffraction intensity distribution is used to incident a broadband beam onto a reference sample via an incident optical path and acquire the broadband partially coherent diffraction intensity distribution of the far field of Fraunhofer diffraction under the reference sample; wherein, the reference sample and the sample to be tested are made of the same optical material.

[0026] The module for acquiring the coherent diffraction intensity distribution of a monochromatic beam is used to use a bandpass filter to monochromatize a broadband beam into a monochromatic beam. The monochromatic beam is then incident on a reference sample via an incident optical path to acquire the coherent diffraction intensity distribution of the far-field monochromatic beam under the reference sample.

[0027] The discrete spectral energy density calculation module is used to solve the discrete spectral energy density by establishing a broadband diffraction multivariate linear regression model by simultaneously solving the intensity distribution of partially coherent diffraction in the far field of Fraunhofer diffraction and the intensity distribution of coherent diffraction in a monochromatic beam under the reference sample.

[0028] The module for establishing the diffraction spectral point diffusion transfer matrix is ​​used to establish the diffraction spectral point diffusion transfer matrix corresponding to a monochromatic beam based on the discrete spectral energy density.

[0029] More preferably, the incident form of the incident light path is: using an aperture device to cut the broadband light beam into a micro-spot and incident it onto the reference sample or the sample to be tested;

[0030] Alternatively, a focusing lens can be used to focus a broadband beam onto a reference sample or the sample to be tested.

[0031] More preferably, the broadband diffraction multiple linear regression model is as follows:

[0032] B0=ω T ·X λ +b

[0033]

[0034] in, For the reference sample in the wavelength matrix λ=[λ0,λ1,λ2,…,λ n The diffraction intensity distribution matrix under monochromatic coherent beam incident conditions; ω represents the spectral weight corresponding to the discrete wavelength λ; b is the system noise error component; AT is the linear transformation transfer function; and the scaling transformation coefficient is... λ c The wavelength of a monochromatic beam; m c B0 represents the coherent diffraction intensity distribution of the monochromatic beam in the far field of Fraunhofer diffraction; B0 represents the coherent diffraction intensity distribution of the broadband far field of Fraunhofer diffraction; λ0, λ1, λ2, ..., λ n The wavelengths of the broadband beam are discrete in the spectrum.

[0035] More preferably, the monochromatic matrix equation for the broadband diffraction image is C T B = C T Cm; where C is the diffusion transfer matrix of the diffraction spectral point corresponding to the monochromatic beam; B is the intensity distribution of the broad-spectral coherent diffraction light under the sample; and m is the intensity distribution of the monochromatic diffraction light.

[0036] More preferably, the coherent diffraction computational imaging algorithm includes a coherent diffraction imaging algorithm, a stacked diffraction imaging algorithm, a Fourier stacked algorithm, and a computational holographic imaging algorithm.

[0037] In summary, the technical solutions conceived by this invention have the following beneficial effects compared with the prior art:

[0038] This invention provides a broadband partially coherent diffraction computational imaging method and system. The method utilizes a multivariate linear regression model of partially coherent broadband diffraction intensity to solve the broadband energy density function, and then uses a broadband diffraction intensity monochromaticization algorithm to monochromate the partially coherent diffraction image. This suppresses the decoherence phenomenon caused by broadband diffraction and improves the quality and convergence robustness of broadband partially coherent diffraction computational imaging.

[0039] This invention provides a broadband partially coherent diffraction computational imaging method and system, comprehensively addressing the key challenges and limitations of current state-of-the-art broadband diffraction imaging, and achieving ultra-wideband partially coherent diffraction imaging without prior knowledge. By monochromating the broadband diffraction data, the coherence of the ultra-wideband partially coherent diffraction signal is greatly improved, achieving full utilization of photon energy across the entire spectral bandwidth. Compared with existing coherent diffraction imaging techniques, the method proposed in this invention improves photon energy utilization efficiency by at least two orders of magnitude, thereby significantly reducing the power requirements of the light source and substantially reducing data acquisition time. Furthermore, the method provided by this invention requires no prior spectral information and overcomes the constraints imposed on dispersive samples. Therefore, the method provided by this invention has basic applicability across any broadband wavelength range, and is particularly suitable for applications in EUV lithography defect detection and soft X-ray lensless imaging. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the broadband partial coherence diffraction computational imaging method provided in an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of a broadband partially coherent diffraction computational imaging system provided in an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram illustrating the verification results of the broadband partially coherent diffraction computational imaging method and system provided in the embodiments of the present invention;

[0043] Marker explanation:

[0044] 201 - Incident light path; 202 - Sample to be tested; 203 - Detector; 301 - Broadband beam; 302 - Monochromatic beam; 303 - Discrete spectral energy density; 304 - Coherent diffraction intensity distribution of the far-field monochromatic beam of Fraunhofer diffraction; 305 - Partial coherent diffraction intensity distribution of the far-field broadband beam of Fraunhofer diffraction; 306 - Monochromatic diffraction intensity distribution; 307 - First reconstructed structure; 308 - Second reconstructed structure; 309 - Third reconstructed result. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0046] This invention provides a broadband partially coherent diffraction computational imaging method and system. It utilizes a multivariate linear regression model of partially coherent broadband diffraction intensity to solve for broadband energy, and then uses a broadband diffraction intensity monochromaticization algorithm to monochromate the partially coherent diffraction image, thereby suppressing the decoherence problem caused by broadband diffraction. This solves the technical problems of existing coherent diffraction imaging technology, such as high requirements for light source coherence, limited spectral bandwidth of broadband partially coherent diffraction methods, and high algorithm solution complexity.

[0047] like Figure 1 and Figure 2 As shown, in one aspect, the present invention provides a broadband partially coherent diffraction computational imaging method, comprising the following steps:

[0048] S101: The broadband beam Λ is incident onto the reference sample through the incident optical path 201, and the near-field wavefront distribution ψ of the incident broadband beam Λ is obtained. Λ The sample light field density distribution function is U0(x,y) and the far-field broadband partial coherent diffraction intensity distribution B0 of the Fraunhofer diffraction, and B0 is received by the detector 203; wherein, the sample light field density distribution function is U0(x,y);

[0049] B0 = {FFT[ψ] Λ (x, y)]} 2

[0050] Where FFT stands for Fast Fourier Transform;

[0051] There are two main types of incident light paths 201: one is to use aperture devices such as micro-apertures to cut the broadband beam into micro-spots and incident them onto the reference sample; the other is to use focusing mirrors, such as Fresnel focusing mirrors and focusing lenses, to converge the broadband beam onto the reference sample.

[0052] The broadband beam to be tested is applicable from the extreme ultraviolet band to the far infrared band; specifically, the applicable bandwidth range in this embodiment is 12nm-15nm in the extreme ultraviolet band, with a spectral bandwidth >22%;

[0053] S102: Use a bandpass filter to monochromaticly filter and modulate the broadband beam into a monochromatic beam λ. c =13.5nm, incident on the reference sample through the incident light path, and the near-field wavefront distribution of the monochromatic incident light is obtained. Fraunhofer diffraction far-field monochromatic beam coherent diffraction intensity distribution m c And it is received by detector 203;

[0054]

[0055] Furthermore, the Fraunhofer diffraction light field models in S101 and S102 need to satisfy the Fraunhofer diffraction approximation conditions, which are generally characterized by the Fresnel number F:

[0056]

[0057] Where a is the size of the incident light spot aperture, such as radius; L is the distance between the sample and the detector;

[0058] S103: Intensity distribution of coherent diffraction of a far-field monochromatic beam by simultaneous Fraunhofer diffraction. c By establishing the intensity distribution B0 of the far-field broadband coherent diffraction of Fraunhofer diffraction, a broadband diffraction multivariate linear regression model is established to solve for the discrete spectral energy density ω(λ).

[0059] More preferably, the broadband diffraction multiple linear regression model is expressed as follows:

[0060] B0=ω T ·X λ +b

[0061] in:

[0062]

[0063] in, For the reference sample in the wavelength matrix λ=[λ0,λ1,λ2,…,λ n The diffraction intensity distribution matrix under the condition of monochromatic coherent beam incident is a known condition; λ0, λ1, λ2, ..., λ h λ represents the discrete wavelength of the broadband beam; ω represents the spectral weight at the corresponding discrete wavelength λ; b is the system noise error component; AT is the linear transform transfer function; and the scaling transform coefficient is... Multiple linear regression models are usually singular matrices and generally do not have a single particular solution. They can be iteratively fitted using gradient descent or least squares methods to obtain the globally optimal approximate solution ω(λ).

[0064] S104: Establish the diffraction spectral point diffusion transfer matrix C based on the discrete spectral energy density ω(λ) output from S103;

[0065] In this embodiment, solving the diffraction spectral point diffusion transfer matrix C in S104 can be understood as the contribution of pixel n of m0 to pixel j of B0, given by the portion of pixel n in the scaling mode at wavelength λ that falls on pixel j, with the scaling factor being λ / λ0 multiplied by the corresponding spectral energy density function ω(λ), summed over the entire spectrum Λ; for diffraction field data of an M*N two-dimensional matrix, the transfer matrix C is a four-dimensional tensor;

[0066] S105: Inject a broadband beam Λ onto any test sample 202 made of the same optical material as the reference sample to obtain the broadband partial coherent diffraction intensity distribution B;

[0067] In this embodiment, the broadband partially coherent diffraction intensity distribution B described in S105 can be read out by translating the sample relative to the incident light path to illuminate the light spot, and using a detector.

[0068] S106: Based on the diffraction spectral point diffusion transfer matrix C output from S104 and the broadband partial coherent diffraction intensity distribution B of the sample under test output from S105, establish the broadband diffraction image monochromatic matrix equation C. T B = C T Cm, calculate the monochromatic diffraction intensity distribution m;

[0069] In this embodiment, the broadband diffraction image monochromaticity matrix equation C described in S106 is... T B = C T Cm is ill-conditioned due to the presence of noise in the actual system, and cannot be solved by direct matrix elimination. It is generally fitted iteratively by conjugate gradient descent or least squares method.

[0070] S107: Substitute the monochromatic diffraction intensity distribution m output from S106 into the coherent diffraction computational imaging algorithm to output the coherent diffraction computational imaging reconstruction result.

[0071] In this embodiment, the coherent diffraction computational imaging algorithm in S107 includes, but is not limited to, coherent diffraction imaging algorithm (CDI), ptychography, Fourier ptychography, and computational holography. It should be noted that the coherent diffraction computational imaging algorithm is a general algorithm in the field of coherent diffraction imaging and can be fully applied to many broadband computational imaging methods.

[0072] On the other hand, the present invention provides a broadband partially coherent diffraction computational imaging system, comprising:

[0073] The monochromatic diffraction intensity distribution calculation module is used to incident a broadband beam 301 onto the sample to be tested. Based on the diffusion transfer matrix of the diffraction spectral point corresponding to the monochromatic beam 302 and the coherent diffraction intensity distribution of the broadband part of the Fraunhofer diffraction far field under the sample to be tested, the monochromatic diffraction image monochromatic matrix equation is established to obtain the monochromatic diffraction intensity distribution 306.

[0074] The reconstruction module of coherent diffraction computational imaging is used to substitute the monochromatic diffraction intensity distribution 306 into the coherent diffraction computational imaging algorithm and output the reconstruction result of coherent diffraction computational imaging.

[0075] The module for acquiring broadband partially coherent diffraction intensity distribution is used to incident a broadband beam onto a reference sample through an incident optical path and acquire the broadband partially coherent diffraction intensity distribution 305 of Fraunhofer diffraction far field under the reference sample.

[0076] The module for acquiring the coherent diffraction intensity distribution of a monochromatic beam is used to use a bandpass filter to monochromatize a broadband beam into a monochromatic beam 302. The monochromatic beam is incident on a reference sample through the incident optical path, and the coherent diffraction intensity distribution of the far-field monochromatic beam under the reference sample is acquired 304.

[0077] The discrete spectral energy density calculation module is used to solve the discrete spectral energy density by establishing a broadband diffraction multivariate linear regression model by simultaneously solving the far-field broadband partial coherent diffraction intensity distribution of Fraunhofer diffraction and the coherent diffraction intensity distribution of monochromatic beam under the reference sample.

[0078] The module for establishing the diffraction spectral point diffusion transfer matrix is ​​used to establish the diffraction spectral point diffusion transfer matrix corresponding to a monochromatic beam based on the discrete spectral energy density.

[0079] More preferably, the incident form of the incident light path is: using an aperture device to cut the broadband light beam into a micro-spot and incident it onto the reference sample or the sample to be tested;

[0080] Alternatively, a focusing lens can be used to focus a broadband beam onto a reference sample or the sample to be tested.

[0081] More preferably, the broadband diffraction multiple linear regression model is as follows:

[0082] B0=ω T ·X λ +b

[0083]

[0084] in, For the reference sample in the wavelength matrix λ=[λ0,λ1,λ2,…,λ n The diffraction intensity distribution matrix under monochromatic coherent beam incident conditions; ω represents the spectral weight corresponding to the discrete wavelength λ; b is the system noise error component; AT is the linear transformation transfer function; and the scaling transformation coefficient is... λ c The wavelength of a monochromatic beam; m c B0 represents the coherent diffraction intensity distribution of the monochromatic beam in the far field of Fraunhofer diffraction; B0 represents the coherent diffraction intensity distribution of the broadband far field of Fraunhofer diffraction; λ0, λ1, λ2, ..., λ n The wavelengths of the broadband beam are discrete in the spectrum.

[0085] More preferably, the monochromatic matrix equation for the broadband diffraction image is C T B = C T Cm; where C is the diffusion transfer matrix of the diffraction spectral point corresponding to the monochromatic beam; B is the intensity distribution of the broad-spectral coherent diffraction light under the sample; and m is the intensity distribution of the monochromatic diffraction light.

[0086] More preferably, the coherent diffraction computational imaging algorithm includes a coherent diffraction imaging algorithm, a stacked diffraction imaging algorithm, a Fourier stacked algorithm, and a computational holographic imaging algorithm.

[0087] Furthermore, such as Figure 3 As shown, in this embodiment, S107 uses a coherent diffraction imaging algorithm for verification; the coherent diffraction intensity distribution m0 output in S102 is solved by coherent diffraction imaging to obtain the first reconstruction structure 307; the broadband partial coherent diffraction intensity distribution B output in S105 is solved by coherent diffraction imaging to obtain the second reconstruction result 308; the monochromatic diffraction intensity distribution m output in S106 is solved by coherent diffraction imaging to obtain the third reconstruction result 309; comparing the second reconstruction result 308 and the third reconstruction result 309, it can be seen that the broadband partial coherent diffraction calculation imaging method and system can significantly improve the quality of the reconstructed image.

[0088] In summary, compared with the prior art, the present invention has the following advantages:

[0089] This invention provides a broadband partially coherent diffraction computational imaging method and system. By using a multivariate linear regression model of partially coherent broadband diffraction intensity to solve the broadband energy density function, and then using a broadband diffraction intensity monochromaticization algorithm to monochromate the partially coherent diffraction image, the decoherence phenomenon caused by broadband diffraction is suppressed, thereby improving the broadband partially coherent diffraction computational imaging quality and convergence robustness.

[0090] The broadband partially coherent diffraction computational imaging method provided by this invention has advantages such as strong algorithm robustness, fast solution, no need for spectral prior information, large spectral bandwidth range, and high imaging quality. It can realize broadband partially coherent diffraction computational imaging in multiple scenes and at wide wavelengths.

[0091] This invention provides a broadband partially coherent diffraction computational imaging method and system, comprehensively addressing the key challenges and limitations of current state-of-the-art broadband diffraction imaging, and achieving ultra-wideband partially coherent diffraction imaging without prior knowledge. By monochromating the broadband diffraction data, the coherence of the ultra-wideband partially coherent diffraction signal is greatly improved, achieving full utilization of photon energy across the entire spectral bandwidth. Compared with existing coherent diffraction imaging techniques, the method proposed in this invention improves photon energy utilization efficiency by at least two orders of magnitude, thereby significantly reducing the power requirements of the light source and substantially reducing data acquisition time. Furthermore, the method provided by this invention requires no prior spectral information and overcomes the constraints imposed on dispersive samples. Therefore, the method provided by this invention has basic applicability across any broadband wavelength range, and is particularly suitable for applications in EUV lithography defect detection and soft X-ray lensless imaging.

[0092] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A wide-spectrum partially-coherent diffractive computational imaging method, characterized in that, Includes the following steps: A broadband beam is incident on the sample to be tested. Based on the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam and the coherent diffraction intensity distribution of the broadband part of the Fraunhofer diffraction far field under the sample to be tested, the monochromatic diffraction image monochromatic matrix equation is established, and the monochromatic diffraction intensity distribution is obtained. Substitute the monochromatic diffraction intensity distribution into the coherent diffraction computational imaging algorithm to output the coherent diffraction computational imaging reconstruction result; The method for obtaining the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam is as follows: A broadband beam is incident on a reference sample through an incident optical path to obtain the coherent diffraction intensity distribution of the broadband far-field Fraunhofer diffraction under the reference sample; wherein, the optical materials of the reference sample and the sample to be tested are the same; A bandpass filter is used to monochromaticly filter and modulate a broadband beam into a monochromatic beam. The monochromatic beam is then incident on a reference sample through the incident optical path to obtain the coherent diffraction intensity distribution of the far-field monochromatic beam of Fraunhofer diffraction under the reference sample. By combining the intensity distribution of partially coherent far-field wide-spectrum Fraunhofer diffraction and the intensity distribution of coherent diffraction of monochromatic beam under the reference sample, a multivariate linear regression model of wide-spectrum diffraction is established to solve the discrete spectral energy density. Establish the diffusion and transfer matrix of the diffraction spectral points corresponding to the monochromatic beam based on the discrete spectral energy density; The broadband diffraction multiple linear regression model is as follows: in, For reference samples in the wavelength matrix The intensity distribution matrix of diffracted light under the condition of monochromatic coherent beam incident; Represents the corresponding discrete wavelength Spectral weights of location; For system noise error components, Let be the linear transformation transfer function, and the scaling transformation coefficients be... ; The wavelength of a monochromatic beam; The intensity distribution of coherent diffraction of a monochromatic beam in the far field of Fraunhofer diffraction; The intensity distribution of the far-field broadband coherent diffraction light in Fraunhofer diffraction; The discrete wavelengths of the broadband beam; The monochromatization matrix equation for the broadband diffraction image is: Where C is the diffusion transfer matrix of the diffraction spectral point corresponding to the monochromatic beam; The intensity distribution of the broad-spectral coherent diffraction light under the sample under test; This represents the monochromatic diffraction intensity distribution.

2. The wide-spectrum partially-coherent diffractive computational imaging method of claim 1, wherein, The incident light path is formed by using an aperture device to cut the broadband light beam into a spot and incident it onto the reference sample or the sample to be tested. Alternatively, a focusing lens can be used to converge a broadband beam of light into a spot and direct it onto the reference sample or the sample to be tested.

3. The broadband partially coherent diffraction computational imaging method according to claim 1, characterized in that, Coherent diffraction computational imaging algorithms include coherent diffraction imaging algorithms, stacked diffraction imaging algorithms, Fourier stacked algorithms, and computational holographic imaging algorithms.

4. A wide-spectrum partially-coherent diffractive computational imaging system, characterized in that, include: The monochromatic diffraction intensity distribution calculation module is used to incident a broadband beam onto the sample to be tested. Based on the diffusion transfer matrix of the diffraction spectral points corresponding to the monochromatic beam and the coherent diffraction intensity distribution of the broadband part of the Fraunhofer diffraction far field under the sample to be tested, the monochromatic diffraction image monochromatic matrix equation is established, and the monochromatic diffraction intensity distribution is obtained. The reconstruction module of coherent diffraction computational imaging is used to substitute the monochromatic diffraction intensity distribution into the coherent diffraction computational imaging algorithm and output the reconstruction result of coherent diffraction computational imaging. The module for acquiring broadband partially coherent diffraction intensity distribution is used to incident a broadband beam onto a reference sample via an incident optical path and acquire the broadband partially coherent diffraction intensity distribution of the far field of Fraunhofer diffraction under the reference sample; wherein, the reference sample and the sample to be tested are made of the same optical material. The module for acquiring the coherent diffraction intensity distribution of a monochromatic beam is used to use a bandpass filter to monochromatize a broadband beam into a monochromatic beam. The monochromatic beam is then incident on a reference sample via an incident optical path to acquire the coherent diffraction intensity distribution of the far-field monochromatic beam under the reference sample. The discrete spectral energy density calculation module is used to solve the discrete spectral energy density by establishing a broadband diffraction multivariate linear regression model by simultaneously solving the intensity distribution of partially coherent diffraction in the far field of Fraunhofer diffraction and the intensity distribution of coherent diffraction in a monochromatic beam under the reference sample. The module for establishing the diffraction spectral point diffusion transfer matrix is ​​used to establish the diffraction spectral point diffusion transfer matrix corresponding to a monochromatic beam based on the discrete spectral energy density. The broadband diffraction multiple linear regression model is as follows: in, For reference samples in the wavelength matrix The intensity distribution matrix of diffracted light under the condition of monochromatic coherent beam incident; Represents the corresponding discrete wavelength Spectral weights of location; For system noise error components, Let be the linear transformation transfer function, and the scaling transformation coefficients be... ; The wavelength of a monochromatic beam; The intensity distribution of coherent diffraction of a monochromatic beam in the far field of Fraunhofer diffraction; The intensity distribution of the far-field broadband coherent diffraction light in Fraunhofer diffraction; The discrete wavelengths of the broadband beam; Wherein, the equation for the monochromatic matrix of the broadband diffraction image is: Where C is the diffusion transfer matrix of the diffraction spectral point corresponding to the monochromatic beam; The intensity distribution of the broad-spectral coherent diffraction light under the sample under test; This represents the monochromatic diffraction intensity distribution.

5. The wide-spectrum partially-coherent diffractive computational imaging system of claim 4, wherein, The incident light path is formed by using an aperture device to cut the broadband light beam into a micro-spot and incident it onto the reference sample or the sample to be tested. Alternatively, a focusing lens can be used to focus a broadband beam onto a reference sample or the sample to be tested.

6. The wide-spectrum partially-coherent diffractive computational imaging system of claim 4, wherein, The coherent diffraction computational imaging algorithm includes a coherent diffraction imaging algorithm, a stacked diffraction imaging algorithm, a Fourier stacked algorithm, and a computational holographic imaging algorithm.

Images