A TDI imaging method and system based on a bayer array image sensor

Abstract

The application provides a TDI imaging method and system based on a Bayer array image sensor, and comprises the following steps: constructing a TDI imaging model based on the Bayer array image sensor according to the working principle of the TDI mode; designing a Bayer array image sensor circuit according to a preset pixel mask strategy, and constructing a coded TDI imaging model based on the Bayer array image sensor; and reconstructing a two-dimensional coded measurement image collected through an image reconstruction algorithm to obtain a three-dimensional color (spectrum) reconstructed image. The Bayer array image sensor circuit is modified, and an image signal processing method is combined, so that the Bayer array sensor can be used for TDI imaging. The TDI imaging model of the Bayer array image sensor for mapping a two-dimensional image to a three-dimensional space is constructed, a color (spectrum) image is reconstructed, and the TDI imaging model has the characteristics of compact structure, high reconstruction quality, good generalization and high robustness.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of remote sensing optical imaging, in particular to a TDI imaging method and system based on a Bayer array image sensor. BACKGROUND

[0002] In modern remote sensing optical imaging, time delay integration (TDI) and video imaging are two mainstream modes, each serving different observation tasks. TDI significantly improves the signal-to-noise ratio and dynamic range of images by multiple exposures and signal accumulation of the same target, and is particularly suitable for mapping in high-speed moving platforms or low-light environments; while video imaging focuses on high-frame-rate real-time monitoring, which can effectively capture dynamic processes such as traffic and disasters. Although many remote sensing satellites carry both TDI and area array sensors to achieve functional complementation, due to the stringent restrictions on satellite size, weight and power consumption (SWaP), the development of a single sensor to support both modes has become an important trend. In this context, different technical paths have been adopted by satellite projects such as SkySat and "Zhu Hai No. 1". SkySat satellites use an innovative CMOS frame camera solution, with the panchromatic sensor responsible for video mode, and in TDI mode, digital technology is used to synchronize orbital speed to integrate and image multispectral images. This design can balance high-resolution panchromatic video and multispectral images, but the system integration is complex. In contrast, the "Zhu Hai No. 1" satellite uses a push-broom Bayer array CMOS sensor, which simulates a virtual linear TDI array by activating only one specific spectral channel in each 2x2 pixel unit. However, this method results in about two-thirds of the pixels not being effectively utilized, sacrificing the original spatial resolution of the sensor and wasting hardware resources. SUMMARY

[0003] The present application proposes a TDI imaging method and system based on a Bayer array image sensor, aiming to realize video imaging and TDI mode imaging using a single Bayer array image sensor. By designing a specific circuit to optimize the direct integration of the Bayer array and the image sensor, and combining an image reconstruction algorithm to obtain multispectral images, the problem of existing single Bayer array image sensors being unable to realize TDI working mode is solved. The technical solution provided by the present application is as follows:

[0004] In a first aspect, a TDI imaging method based on a Bayer array image sensor is provided, comprising the following steps:

[0005] S1. According to the working principle of TDI mode, a TDI imaging mode of the Bayer array image sensor is constructed;

[0006] S2. Based on a preset pixel masking strategy, design the Bayer array image sensor circuit and construct a coded TDI imaging model based on the Bayer array image sensor. In TDI mode, the Bayer array image sensor has K TDI levels and an array size of N×N, where K≤N. Each pixel in the sensor corresponds one-to-one with a Bayer array pixel. Define a Cartesian coordinate system with the push-broom direction of the standard Bayer array sensor as the x-axis and the orthogonal direction as the y-axis. During imaging, scene radiation is focused onto the sensor surface by the objective lens system and generates pixel-level voltage signals through photoelectric conversion. When the sensor translates one pixel unit relative to the scene along the x-axis, the driving circuit, based on a preset pixel masking strategy, sets the output of a specific pixel to zero, preventing it from participating in TDI accumulation. According to the TDI integration timing, the pixel signals processed by the preset masking strategy are accumulated. This process superimposes pixel signals corresponding to the same ground sampling point at multiple TDI levels, outputting a one-dimensional measurement vector with spectral information aliasing. By caching and stitching the continuously acquired one-dimensional measurement vectors, a two-dimensional coded measurement image of the push-broom is obtained.

[0007] S3. The acquired two-dimensional coded measurement image is reconstructed using an image reconstruction algorithm to obtain a three-dimensional reconstructed spectral image.

[0008] The image sensor contains M×M Bayer array units, where each Bayer array unit contains k×k Bayer array pixels. The area of ​​the image sensor is kM×kM, i.e., N=kM. The Bayer array pixels are coated with thin films with different spectral transmittances to acquire spectral images of different wavelengths.

[0009] The TDI imaging model for the Bayer array image sensor is expressed as follows:

[0010]

[0011] in, This represents the dot product operation. An image is obtained by superimposing pixel signals from multiple TDI levels of a Bayer array image sensor corresponding to the same ground sampling point. , The desired multispectral image The vectorized representation of, It is the sensing matrix of the Bayer array. It is the spectral matrix of the Bayer array.

[0012] The expression for the perception matrix is:

[0013]

[0014] The entire perception matrix The coverage area includes all pixels on the Bayer array image sensor array; According to the pixel masking strategy, the spectral band corresponding to the generated voltage that does not participate in TDI accumulation is set to "0", while the value of the generated voltage that participates in TDI accumulation is "1".

[0015] The expression for the spectral matrix is:

[0016] ;

[0017] It is a narrowband spectral transmittance function, taking values ​​of "1" or "0", and has a total of indivual Let the value be 0. ,when When the narrowband spectral transmittance function is expressed as:

[0018]

[0019] when When the narrowband spectral transmittance function is expressed as:

[0020] .

[0021] The vectorized representation of the desired multispectral image is:

[0022]

[0023] Where, in the formula This represents the center wavelength contained in the j-th pixel of the image cached in TDI mode. The value to be rebuilt.

[0024] The N×1 set of measurements obtained by push-broom acquisition are used to sequentially extract the center wavelength from each image pixel. The measured values ​​are formed into a two-dimensional vector, resulting in 1×k. 2 The N two-dimensional vectors are then stacked to form a three-dimensional vector, resulting in an N×1×k vector. 2 Three-dimensional multispectral images.

[0025] When reconstructing an image, each pixel of the three-dimensional multispectral image is first mapped into a three-dimensional space, denoted as 1×1×k. 2 Then, the 3D image is used as input to the reconstruction model; the reconstruction is transformed into a mathematical convergence problem, expressed as:

[0026]

[0027] in, Indicates the reconstruction result. This is a three-dimensional matrix representing the input for reconstruction. denotes a norm, and is a positive integer;

[0028] The reconstructed result is N x 1 x k 2 The three-dimensional reconstruction image of the three-dimensional reconstruction image is shown in the following table: to The corresponding spectral image.

[0029] In a second aspect, a TDI imaging system based on a Bayer array image sensor includes a circuit design module, an integrated design module, a data acquisition system, and an image reconstruction module.

[0030] The circuit design module is realized by designing a Bayer array image sensor acquisition circuit to realize data acquisition of the Bayer array sensor in TDI mode. The integrated design module is to integrate the circuit with the existing Bayer array image sensor to ensure that the hardware and circuit are compatible. The data acquisition system is to control the imaging device to perform image acquisition process, and the acquisition circuit is designed to realize data acquisition of the Bayer array sensor in TDI mode according to the preset pixel mask strategy. The image reconstruction module is to use open source or customized algorithm to reconstruct the two-dimensional spectral information mixed image into a color spectral image.

[0031] Compared with the prior art, the present application has the following beneficial effects: by modifying the image sensor, integrating the Bayer array on the image sensor, designing the corresponding circuit, and reconstructing the two-dimensional spectral information mixed image obtained by scanning into a color (spectral) image, TDI mode imaging is realized on a single Bayer array image sensor. BRIEF DESCRIPTION OF DRAWINGS

[0032] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, which together with the embodiments of the application, serve to explain the application, and do not constitute a limitation on the application. In the drawings:

[0033] Figure 1 is the overall flowchart of the TDI imaging method provided by the present application;

[0034] Figure 2 is the structure block diagram and the Bayer array design diagram in the TDI imaging method provided by the present application;

[0035] Figure 3 is the Bayer array pixel mask strategy and two-dimensional space mapping to three-dimensional space diagram in the TDI imaging method provided by the present application;

[0036] Figure 4 is the structure diagram of the TDI imaging system provided by the present application;

[0037] Figure 5 is a simulation result diagram of the TDI imaging system provided by the present application. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work belong to the protection scope of the present application.

[0039] In order to make the above-mentioned purposes, features and effects of the present application more apparent and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0040] Embodiment 1, Reference Figures 1-3 For the first embodiment of the present application, a TDI imaging method based on a Bayer array image sensor is provided, Figure 1 A flowchart of the method is provided, including the following steps.

[0041] S1, according to the working principle of the TDI mode, a TDI imaging mode of the Bayer array image sensor is constructed, Figure 2 A structural block diagram of the TDI imaging method and a Bayer array design diagram are provided.

[0042] Specifically, the entire image sensor contains MxM Bayer array units, wherein a single Bayer array unit contains kxk Bayer array pixels, the surface array scale of the image sensor is kMxkM, that is, N=kM; the Bayer array pixels are respectively plated with thin films with different spectral transmittances, which are used to collect spectral images of different wavebands.

[0043] Under the TDI working mode of the Bayer array image sensor, the TDI series is K, K≤N, and each pixel in the image sensor corresponds to a Bayer array pixel. First, a Cartesian coordinate system is defined with the push-broom direction of a standard Bayer array sensor as the x-axis and the orthogonal direction as the y-axis. In the imaging process, the scene radiation is focused on the sensor surface through the objective system, and the pixel-level voltage signal is generated through photoelectric conversion. When the sensor is translated by one pixel unit along the x-axis direction relative to the scene, the driving circuit will zero the output of the specific pixel according to a preset pixel mask strategy. Subsequently, the pixel signal processed by the preset mask strategy is accumulated according to the TDI integration timing. This process superimposes the pixel signals of the corresponding ground sampling points on multiple TDI levels, thereby outputting a one-dimensional measurement vector with spectral information superimposed. Finally, by buffering and splicing the continuously acquired one-dimensional measurement vectors, a two-dimensional raw measurement image of the push-broom can be obtained

[0044] S2, according to the preset pixel mask strategy, design the sensor circuit, and construct the coding TDI imaging model based on the Bayer array image sensor.

[0045] Specifically, in the process of pushing through the ground by one pixel, the first kMxk Bayer array pixels along the x-axis direction are set to zero according to the preset strategy, and M times are repeated, that is, the pixel mask strategy of the previous Bayer array unit along the x-axis direction is repeated, the output of the specific pixel is set to zero, and a group of measurement values are output after pushing through the ground by one pixel in TDI mode. When the next pixel is pushed, the zero processing of the Bayer array pixels can be selected again, but a group of measurement values are arranged in order into a one-dimensional vector y according to the above rules, which can be represented as:

[0046]

[0047] Through the cumulative integration in TDI mode, each pixel value in a group of measurement values can be represented as the weighted sum of the corresponding spectral signal, and the jth measurement value can be represented as:

[0048]

[0049] wherein represents the jth measurement value in a group of measurement values. is the zero processing of the output of the specific pixel according to the preset strategy, and the output of other pixels is taken as "1". is a narrowband spectral transmittance function, which takes the value of "1" or "0", and there are The value is 0. represents the to-be-reconstructed value corresponding to the center wavelength .Let ;

[0050] When , the narrowband spectral transmittance function is represented as:

[0051]

[0052] When , the narrowband spectral transmittance function is represented as:

[0053] .

[0054] For a group of measurement values of one size Nx1 obtained by pushing and collecting, the measurement values with the center wavelength are extracted from each image pixel in turn to form a two-dimensional vector, and a two-dimensional vector of 1xk 2 is obtained; then N two-dimensional vectors are stacked to form a three-dimensional vector, and a size of Nx1xk 2Three-dimensional multispectral images.

[0055] The measurement value of the j-th pixel in a N×1 set of measurements obtained by push-broom acquisition. The expression is:

[0056]

[0057] in, It is the value of the j-th pixel relative to the i-th spectral band, determined by a preset pixel masking strategy. It takes the value of "0" for a specific pixel and "1" for other pixels. It is a narrowband spectral transmittance function, taking values ​​of "1" or "0", and has a total of indivual The value is 0. These are the multispectral image pixel values ​​that the Bayer array image sensor expects to obtain; let ;

[0058] when When the narrowband spectral transmittance function is expressed as:

[0059]

[0060] when When the narrowband spectral transmittance function is expressed as:

[0061]

[0062] The TDI imaging model for the Bayer array image sensor is expressed as follows:

[0063]

[0064] in, This represents the dot product operation. The image is obtained by scanning one pixel of the ground using a Bayer array image sensor. , The desired multispectral image The vectorized representation of, It is the sensing matrix of the Bayer array. It is the spectral matrix of the Bayer array.

[0065] The perception matrix is ​​expressed as:

[0066]

[0067] in, The coverage area includes all pixels on the Bayer array image sensor's array.

[0068] The spectral matrix is ​​expressed as:

[0069] .

[0070] The vectorized representation of the desired multispectral image is:

[0071]

[0072] wherein in the formula represents the value to be reconstructed of the corresponding center wavelength contained in the jth pixel in the image obtained by caching in the TDI mode. .

[0073] S3, the image collected by the TDI imaging method based on the Bayer array image sensor is reconstructed by an image reconstruction algorithm to obtain a three-dimensional reconstructed spectral image.

[0074] When reconstructing the image, first map each pixel of the image to a three-dimensional space, denoted as 1x1xk 2 , and then take the three-dimensional image Y as the input of the reconstruction model; convert the reconstruction into a mathematical convergence problem, and the expression is:

[0075]

[0076] wherein, represents the reconstruction result, is a three-dimensional matrix, representing the input of the reconstruction, represents a norm, and are positive integers;

[0077] The three-dimensional reconstructed image after reconstruction is Nx1xk 2 , and the corresponding center wavelength from bottom to top along the Z axis is to corresponding spectral image.

[0078] Figure 3 , the Bayer array pixel mask strategy and the two-dimensional space to three-dimensional space mapping diagram are provided. Specifically, Figure 3 (a) is a schematic diagram of a Bayer array, when k=2, M=4, the Bayer array image sensor surface array size is 8x8, the Bayer array image sensor is the current common RGGB mode, the square surrounded by the black solid line in the figure represents a 2x2 Bayer array unit, each Bayer array unit contains 4 black dotted line frame small squares, from left to right and from top to bottom, they are "R", "G", "G", "B", and the x direction is the push scanning direction. Figure 3 (b) is a schematic diagram of a pixel mask strategy, for Figure 3(a) The Bayer array image sensor shown is designed with a preset strategy for pixel mask strategy for the first 8x2 Bayer array pixels along the x-axis direction, and repeated 4 times, that is, the pixel mask strategy of the previous Bayer array unit is repeated along the x-axis direction, and the output of the black square pixels in the figure is zeroed. Figure 3 (c) Schematic diagram for mapping two-dimensional spectral aliasing image to three-dimensional space.

[0079] In order to verify the beneficial effects of the present application, the existing traditional method and the method of the present application are respectively experimented by economic benefit calculation and simulation experiment.

[0080] Reference Figure 5are part of the results of a set of natural images reconstructed by the TDI imaging method simulation experiment of the present application. Among them, (a1), (a2), (a3), (a4) are the original color (spectrum) images and the corresponding "R", "G", "B" channel images, (b1), (b2), (b3), (b4) are the images and the corresponding "R", "G", "B" channel images obtained after reconstruction by the method of the present application, the PSNR of (a1) and (b1) is 37.63dB, the SSIM is 0.990, the average color difference under the CIEDE2000 standard is 1.0535, the PSNR of (a2) and (b2) is 41.43dB, the SSIM is 0.932, the PSNR of (a3) and (b3) is 44.37dB, the SSIM is 0.958, the PSNR of (a4) and (b4) is 41.96dB, the SSIM is 0.990; (c1), (c2), (c3), (c4) are the original color (spectrum) images and the corresponding "R", "G", "B" channel images, (d1), (d2), (d3), (d4) are the images and the corresponding "R", "G", "B" channel images obtained after reconstruction by the method of the present application, the PSNR of (c1) and (d1) is 36.46dB, the SSIM is 0.994, the average color difference under the CIEDE2000 standard is 1.1425, the PSNR of (c2) and (d2) is 40.79dB, the SSIM is 0.920, the PSNR of (c3) and (d3) is 42.86dB, the SSIM is 0.954, the PSNR of (c4) and (d4) is 40.43dB, the SSIM is 0.930; (e1), (e2), (e3), (e4) are the original color (spectrum) images and the corresponding "R", "G", "B" channel images, (f1), (f2), (f3), (f4) are the images and the corresponding "R", "G", "B" channel images obtained after reconstruction by the method of the present application, the PSNR of (e1) and (f1) is 37.53dB, the SSIM is 0.993, the average color difference under the CIEDE2000 standard is 1.0348, the PSNR of (e2) and (f2) is 41.76dB, the SSIM is 0.930, the PSNR of (e3) and (f3) is 44.25dB, the SSIM is 0.963, the PSNR of (e4) and (f4) is 41.41dB, the SSIM is 0.940.

[0081] Example 2: Refer to Figure 4 As a second embodiment of the present application, a TDI imaging system based on a Bayer array image sensor is provided, including a circuit design module, an integrated design module, a data acquisition system, an image reconstruction module.

[0082] The circuit design module is to realize data acquisition of the Bayer array sensor in TDI mode through design of a Bayer array image sensor acquisition circuit; the integrated design module is to integrate the circuit with an existing Bayer array image sensor to ensure adaptation of the hardware and the circuit; the data acquisition system is to control an imaging device to perform image acquisition, and the acquisition circuit is designed to realize data acquisition of the Bayer array sensor in TDI mode according to a preset pixel mask strategy; and the image reconstruction module is to use an open source or customized algorithm to reconstruct a two-dimensional spectral information aliasing image into a color image.

[0083] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0084] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered a list of executable instructions for implementing logic functions, and can be specifically embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other system that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions, or in conjunction with these instruction execution systems, apparatuses, or devices. For the purpose of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by an instruction execution system, apparatus, or device, or in conjunction with these instruction execution systems, apparatuses, or devices.

[0085] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for instance via an optical scanner, then compiled, interpreted, or otherwise processed, using suitable tools, and then stored in a computer memory.

[0086] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware which are stored in memory and executed by a suitable instruction execution system.

[0087] The above descriptions are only the preferred embodiment of the application, not intended to limit the application and although the application has been described in detail by referring to the foregoing embodiments, those skilled in the art could still make modifications to the technical solutions described in the foregoing embodiments or make equivalent replacements to some of the technical features, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A TDI imaging method based on a Bayer array image sensor, characterized in that, Includes the following steps: S1. Based on the working principle of TDI mode, construct the TDI imaging mode of Bayer array image sensor; S2. Based on a preset pixel mask strategy, design the Bayer array image sensor circuit and construct an coded TDI imaging model based on the Bayer array image sensor. In TDI mode, the Bayer array image sensor has a TDI level of K and contains M×M Bayer array units, where each Bayer array unit contains k×k Bayer array pixels. The area of ​​the image sensor is kM×kM, i.e., N=kM, K≤N. Each pixel in the sensor corresponds one-to-one with a Bayer array pixel. Define a Cartesian coordinate system with the push-broom direction of the standard Bayer array sensor as the x-axis and the orthogonal direction as the y-axis. During imaging, scene radiation is focused onto the sensor surface by the objective lens system and generates pixel-level voltage signals through photoelectric conversion. When the sensor translates one pixel unit relative to the scene along the x-axis, the driving circuit sets the output of the pixel set in the pixel mask strategy to zero according to a preset pixel mask strategy, and the pixel does not participate in TDI accumulation. According to the TDI integration time sequence, the pixel signals processed by the preset masking strategy are accumulated. This process superimposes the pixel signals corresponding to the same ground sampling point at multiple TDI levels and outputs a one-dimensional measurement vector with spectral information aliasing. By caching and stitching the continuously acquired one-dimensional measurement vectors, a two-dimensional coded measurement image of pushbroom is obtained. S3. The acquired two-dimensional coded measurement image is reconstructed using an image reconstruction algorithm to obtain a three-dimensional reconstructed spectral image.

2. The TDI imaging method based on a Bayer array image sensor according to claim 1, characterized in that, The image sensor contains M×M Bayer array units, where each Bayer array unit contains k×k Bayer array pixels. The area of ​​the image sensor is kM×kM, i.e., N=kM. The Bayer array pixels are coated with thin films with different spectral transmittances to acquire spectral images of different wavelengths.

3. The TDI imaging method based on a Bayer array image sensor according to claim 1, characterized in that, The TDI imaging model for the Bayer array image sensor is expressed as follows: ； in, This represents the dot product operation. It is an image obtained by superimposing pixel signals corresponding to the same ground sampling point on multiple TDI levels of the Bayer array image sensor. The desired multispectral image The vectorized representation of, It is the sensing matrix of the Bayer array. It is the spectral matrix of the Bayer array.

4. The TDI imaging method based on a Bayer array image sensor according to claim 3, characterized in that, The expression for the perception matrix is: ； The entire perception matrix The coverage area includes all pixels on the Bayer array image sensor array; This indicates the corresponding center wavelength contained in the j-th pixel. According to the pixel masking strategy, the spectral band corresponding to the generated voltage that does not participate in TDI accumulation is set to "0", and the generated voltage that participates in TDI accumulation is set to "1".

5. The TDI imaging method based on a Bayer array image sensor according to claim 3, characterized in that, The expression for the spectral matrix is: ； This indicates the corresponding center wavelength contained in the j-th pixel. The narrowband spectral transmittance function takes values ​​of "1" or "0", and has a total of indivual Let the value be 0. ,when When the narrowband spectral transmittance function is expressed as: ； when When the narrowband spectral transmittance function is expressed as: 。 6. The TDI imaging method based on a Bayer array image sensor according to claim 3, characterized in that, The vectorized representation of the desired multispectral image is: ； Where, in the formula This represents the center wavelength contained in the j-th pixel of the image cached in TDI mode. The value to be rebuilt.

7. The TDI imaging method based on a Bayer array image sensor according to claim 3, characterized in that, The N×1 set of measurements obtained by push-broom acquisition are used to sequentially extract the center wavelength from each image pixel. The measured values ​​are formed into a two-dimensional vector, resulting in 1×k. 2 The N two-dimensional vectors are then stacked to form a three-dimensional vector, resulting in an N×1×k vector. 2 Three-dimensional multispectral images.

8. The TDI imaging method based on a Bayer array image sensor according to claim 7, characterized in that, When reconstructing an image, each pixel of the three-dimensional multispectral image is first mapped into a three-dimensional space, denoted as 1×1×k. 2 Then, the 3D image is used as input to the reconstruction model; the reconstruction is transformed into a mathematical convergence problem, expressed as: ； in, Indicates the reconstruction result. This is a three-dimensional matrix representing the input for reconstruction. Represents the norm, and It is a positive integer; The reconstructed result is N×1×k 2 The three-dimensional reconstructed image, arranged from bottom to top along the Z-axis, shows the corresponding center bands. to The corresponding spectral image.

9. A TDI imaging system based on a Bayer array image sensor, characterized in that, The system includes a circuit design module, an integrated design module, a data acquisition system, and an image reconstruction module. The above modules are described in any one of claims 1-8 as a TDI imaging method based on a Bayer array image sensor to achieve imaging.

Images