Hyperspectral linear array TDI image sensor and image processing method

Abstract

The invention discloses a hyperspectral linear array TDI image sensor and an image processing method, and the hyperspectral linear array TDI image sensor comprises a substrate which comprises a pixel region; a plurality of pixel arrays, the plurality of pixel arrays are located on the pixel area, and the plurality of pixel arrays are sequentially arranged along a first direction; the array substrate comprises a plurality of pixel arrays, the super-structure surface arrays are located on the pixel arrays, each super-structure surface array comprises a plurality of super-surface micro-nano structures, and the super-surface micro-nano structures are in one-to-one correspondence with the pixel arrays. The metasurface micro-nano structure is located on the pixel array, the metasurface micro-nano structure carries out broadband modulation on received incident light, the brightness of images of all channels received by the pixel array and subjected to broadband modulation through the metasurface micro-nano structure is high, and the signal-to-noise ratio of the images can be improved; in addition, the hyperspectral linear array TDI image sensor does not need to use an optical filter, so that the production and design costs are reduced.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor manufacturing technology, and in particular to a hyperspectral linear array TDI image sensor and an image processing method. Background Art

[0002] Linear array TDI (Time-Delay Integrated) image sensor is an image sensor that uses TDI technology to expose the same target multiple times and accumulate the exposure results to improve the signal-to-noise ratio and dynamic range.

[0003] The main features of linear array TDI image sensors are high sensitivity, high resolution, low power consumption, high system integration, low readout noise, and high scanning speed. It is widely used in imaging fast objects in dim light, such as industrial assembly line inspection, food inspection, biological inspection, 3C production inspection and other fields. Compared with black and white cameras and traditional color cameras, multispectral (hyperspectral) imaging technology can provide rich target information and greatly improve the recognition efficiency and detection efficiency of the detected objects, which has put the development of multispectral (hyperspectral) linear array TDI image sensors on the fast track.

[0004] However, current hyperspectral linear array TDI image sensors still have some shortcomings. Summary of the invention

[0005] The technical problem solved by the present invention is how to improve the hyperspectral linear array TDI image sensor, effectively improve the signal-to-noise ratio of the image, and restore the full-spectrum image information within a set spectral range from the acquired original data.

[0006] To solve the above technical problems, an embodiment of the present invention provides a hyperspectral linear array TDI image sensor, comprising: a substrate, the substrate comprising a pixel area; a plurality of pixel arrays, the plurality of pixel arrays being located on the pixel area, and the plurality of pixel arrays being arranged sequentially along a first direction; a metasurface array, the metasurface array being located on the plurality of pixel arrays, the metasurface array comprising a plurality of metasurface micro-nano structures, and the metasurface micro-nano structures corresponding one-to-one to the pixel arrays.

[0007] Optionally, the material of the multiple supersurface micro-nano structures is the same.

[0008] Optionally, the material of the supersurface micro-nano structure includes but is not limited to organic matter and silicon oxide.

[0009] Optionally, among the multiple super-surface micro-nano structures, at least two of the super-surface micro-nano structures are made of different materials.

[0010] Optionally, the metasurface micro-nanostructure performs broadband modulation on the received incident light to obtain modulated light; and the corresponding pixel array receives the modulated light.

[0011] Optionally, the structure of the metasurface micro-nanostructure in the metasurface array is determined based on the spectrum of the incident light and the spectral line band collected by the corresponding pixel array.

[0012] Optionally, the metasurface micro-nanostructures in the metasurface array include: modulators periodically distributed in a plane parallel to the substrate surface.

[0013] Optionally, the cross-sectional shape of the modulation element includes but is not limited to: circular, star-shaped, and circular hole.

[0014] Optionally, the super-surface micro-nano structure is in direct contact with the corresponding pixel array.

[0015] Optionally, on the surface of the substrate, the projection of the pixel array corresponding to the super-surface micro-nano structure is located within the range of the projection of the super-surface micro-nano structure.

[0016] Optionally, the number of the pixel arrays is greater than 4.

[0017] Optionally, it also includes: a reconstruction module, and the reconstruction module is configured as: an acquisition unit and a calculation unit.

[0018] Optionally, the reconstruction module is suitable for performing spectral reconstruction on the image to be processed obtained by using the image sensor to obtain spectral information; wherein the acquisition unit is suitable for obtaining the image to be processed based on the received incident light; the operation unit is suitable for using a neural network algorithm to perform operations on the obtained image to be processed, and obtain spectral information based on the result of the operation.

[0019] Correspondingly, an embodiment of the present invention further provides an image processing method, comprising: acquiring an image to be processed using the hyperspectral linear array TDI image sensor as described in any one of the above items; and obtaining spectral information based on the image to be processed.

[0020] Optionally, in the step of obtaining spectral information, spectral information within a preset spectral range is obtained based on the image to be processed.

[0021] Optionally, in the step of obtaining spectral information, the spectral information is obtained based on the image to be processed and the structure of the metasurface array.

[0022] Optionally, in the step of obtaining spectral information, the spectral information is obtained through a neural network algorithm based on the image to be processed and the structure of the metasurface array.

[0023] Compared with the prior art, the technical solution of the embodiment of the present invention has the following beneficial effects:

[0024] In the hyperspectral linear array TDI image sensor of the technical solution of the present invention, the metasurface micro-nano structure is located on the pixel array, and the metasurface micro-nano structure is used to modulate the received incident light. The metasurface micro-nano structure does not filter out the incident light, so that the brightness of the image modulated by the metasurface micro-nano structure received by the pixel array is higher, which is beneficial to improving the signal-to-noise ratio of the image; in addition, the hyperspectral linear array TDI image sensor of this solution does not require the use of filters, and the production and design costs of precision filters are high. The hyperspectral linear array TDI image sensor of this solution effectively reduces the cost of the process.

[0025] In an optional solution of the present invention, the metasurface micro-nano structure is in direct contact with the corresponding pixel array. That is, there is no other structure, such as a filter, between the metasurface micro-nano structure and the pixel array. The incident light received is only broadband modulated according to the metasurface micro-nano structure to obtain modulated light, and the corresponding pixel array receives the modulated light. The production and design costs of precision filters are high, and the hyperspectral linear array TDI image sensor of this solution effectively reduces the cost of the process.

[0026] In the image processing method of the technical solution of the present invention, a hyperspectral linear array TDI image sensor with a super-surface micro-nano structure is used to modulate the received incident light to obtain an image to be processed, thereby obtaining spectral information. The spectral information obtained by the method is not restricted by the number of filters, and full-spectrum image information can be obtained, and the spectral restoration accuracy is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 FIG. 4 is a schematic diagram of a hyperspectral linear array TDI image sensor according to a prior art embodiment.

[0028] Figure 2 It is a schematic diagram of a scene in a prior art embodiment;

[0029] Figure 3 It is a schematic diagram of a channel image output by a hyperspectral linear array TDI image sensor according to a prior art embodiment;

[0030] Figure 4 is a schematic diagram of a hyperspectral linear array TDI image sensor according to an embodiment of the present invention;

[0031] Figure 5 is a schematic diagram of a scene according to an embodiment of the present invention;

[0032] Figure 6 It is a schematic diagram of a spectral image having a number greater than the number of channels, which is calculated based on the image data acquired from each channel according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] As can be seen from the background technology, the hyperspectral linear array TDI image sensor in the prior art still has some shortcomings. The reasons for the shortcomings are now analyzed.

[0034] Figure 1 It is a hyperspectral linear array TDI image sensor of an embodiment of the prior art, comprising: a substrate 100, wherein the substrate 100 comprises a pixel area; a plurality of pixel arrays 101, wherein the plurality of pixel arrays 101 are located on the pixel area, and the plurality of pixel arrays 101 are sequentially arranged along a first direction X; a filter 102, wherein the filter 102 is located on the plurality of pixel arrays 101, and the filter 102 comprises a plurality of filter arrays 103, wherein each filter array 103 is located on each pixel array 101, and the filter arrays 103 correspond to the pixel arrays 101 one by one.

[0035] Specifically, the number of the pixel arrays 101 is 8, and correspondingly, the number of the filter arrays 103 is also 8.

[0036] The hyperspectral linear array TDI image sensor adopts the principle of narrow-band filtering to filter the incident light of a wide spectrum into a plurality of narrow-spectrum lights of interest. The light flux transmitted by each spectrum is generally lower than 20% of the light flux of the incident light, or even lower, resulting in lower image brightness. Moreover, if the spectral information of more spectrum bands is to be expanded, more pixel arrays 101 and filter arrays 103 need to be added. Different applications are concerned with different spectral bands. If a variety of needs are to be met, a large number of filters 102 with different bands and different numbers of spectrum bands need to be designed and produced. However, the production and design costs of precision filters 102 are high.

[0037] In addition, the traditional narrow-band filter image sensor can only obtain a limited number of spectral images, and cannot effectively image images outside the passband range. Figures 1 to 3 , when the number of pixel arrays 101 is 8, the number of filter arrays 103 is also 8, the traditional narrow-band filtering multi-spectral sensor can only obtain spectral information equal to the number of channels, that is, using Figure 1 The narrow-band filter type image sensor shown can only restore channel images of the same number as the number of filter arrays 103, and loses the information of scenes with some wavelengths.

[0038] To solve the above technical problems, the technical solution of the present invention provides a hyperspectral linear array TDI image sensor, including: a substrate, the substrate includes a pixel region; a plurality of pixel arrays, the plurality of pixel arrays are located on the pixel region, and the plurality of pixel arrays are arranged in sequence along a first direction; a metasurface array, the metasurface array is located on the plurality of pixel arrays, the metasurface array includes a plurality of metasurface micro-nano structures, and the metasurface micro-nano structures correspond to the pixel arrays one by one.

[0039] In the hyperspectral linear array TDI image sensor of the technical solution of the present invention, the metasurface micro-nano structure is located on the pixel array, and the metasurface micro-nano structure is used to modulate the received incident light. The metasurface micro-nano structure does not filter out the incident light, so that the brightness of the image received by the pixel array after being modulated by the metasurface micro-nano structure is relatively high, which is beneficial to improving the signal-to-noise ratio of the image; in addition, the hyperspectral linear array TDI image sensor of this solution does not need to use a filter, and the production and design costs of precise filters are high. The hyperspectral linear array TDI image sensor of this solution effectively reduces the process cost.

[0040] In order to make the above objects, features and beneficial effects of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the accompanying drawings.

[0041] Please refer to Figure 4 , the hyperspectral linear array TDI image sensor includes: a substrate 200, and the substrate 200 includes a pixel region.

[0042] The pixel region of the substrate 200 provides structural support for the pixel array 201.

[0043] Please refer to Figure 4 , the hyperspectral linear array TDI image sensor includes: a plurality of pixel arrays 201, the plurality of pixel arrays 201 are located on the pixel region. Specifically, in some embodiments of the present invention, the plurality of pixel arrays 201 are arranged in sequence along the first direction X.

[0044] Each pixel array 201 is composed of pixels with a certain integration level. The level of the pixel array 201 is set to different values according to the application field, and the level of the pixel array 201 ≥ 2 levels.

[0045] The number of the pixel arrays 201 is greater than 4. Specifically, in some embodiments of the present invention, the number of the pixel arrays 201 is 8. In other embodiments, the number of the pixel arrays may also be 12 or 16 or other numbers.

[0046] Please refer to Figure 4The hyperspectral linear array TDI image sensor includes: a metasurface array 202, wherein the metasurface array 202 is located on a plurality of pixel arrays 201, and the metasurface array 202 includes a plurality of metasurface micro-nano structures 203, and the metasurface micro-nano structures 203 correspond one-to-one to the pixel arrays 201.

[0047] The super-surface micro-nano structure 203 is used to modulate incident light.

[0048] The metasurface micro-nanostructure 203 modulates the incident light and does not filter out the light, so that the light intensity obtained by each pixel array 201 is higher and the image brightness is higher, which is beneficial to improve the signal-to-noise ratio of the image. Based on the above advantages of wide spectrum modulation, the hyperspectral linear array TDI image sensor of the present invention can use a lower order or work at a higher line frequency, which can effectively reduce the cost of the hyperspectral linear array TDI image sensor.

[0049] The metasurface array 202 includes a plurality of metasurface micro-nano structures 203, and the metasurface micro-nano structures 203 correspond to the pixel array 201 one by one, that is, the number of the metasurface micro-nano structures 203 is the same as the number of the pixel array 201, and one metasurface micro-nano structure 203 corresponds to one pixel array 201. Specifically, in some embodiments of the present invention, the number of the pixel array 201 is 8, and accordingly, the number of the metasurface micro-nano structures 203 is also 8.

[0050] Specifically, on the surface of the substrate 200, the projection of the pixel array 201 corresponding to the super-surface micro-nano structure 203 is located within the projection range of the super-surface micro-nano structure 203. In order to ensure that the super-surface micro-nano structure 203 can completely cover the photosensitive surface, the projection of the super-surface micro-nano structure 203 and the projection of the pixel array 201 corresponding to the super-surface micro-nano structure 203 should be in an inclusive relationship, that is, the projections of the pixel arrays 201 corresponding to the super-surface micro-nano structure 203 are respectively located within the projection range of each super-surface micro-nano structure 203.

[0051] The metasurface array 202 includes a plurality of metasurface micro-nano structures 203. The structures of the plurality of metasurface micro-nano structures 203 of the metasurface array 202 are different. The metasurface micro-nano structures 203 of different structures can perform different response modulations on the broadband spectrum of the bandwidth range of interest, so as to output modulated light corresponding to each metasurface micro-nano structure 203, and the pixel array 201 corresponding to each metasurface micro-nano structure 203 receives the modulated light. For example, in some embodiments of the present invention, the metasurface array 202 includes 8 metasurface micro-nano structures 203 of different structures, and the 8 metasurface micro-nano structures 203 perform different response modulations on the spectrum of the incident light.

[0052] Specifically, in some embodiments of the present invention, the structure of the metasurface micro-nanostructure 203 in the metasurface array 202 is determined based on the spectrum of the incident light and the spectral line band collected by the corresponding pixel array 201.

[0053] Specifically, in some embodiments of the present invention, in order to ensure the accuracy of spectral reconstruction and restoration, a large number of simulation design tests are carried out on 8 types of metasurface micro-nano structures 203, and spectral reconstruction is performed using a neural network algorithm to verify the rationality of the spectral response curve of the designed metasurface micro-nano structure 203.

[0054] For most applications, a metasurface array (i.e., an 8-channel architecture) having 8 metasurface micro-nano structures 203 can be used to reconstruct spectral information that meets the spectral restoration accuracy requirements. Currently, the spectral restoration accuracy of TDI image sensors using this 8-channel architecture can reach nanometer level or even higher accuracy.

[0055] In other embodiments, it also includes: expanding the spectral channel and redesigning the metasurface micro-nano structure to improve the spectral restoration accuracy of the hyperspectral linear array TDI image sensor, that is, increasing the number of pixel arrays and the number of metasurface micro-nano structures to improve the spectral restoration accuracy of the hyperspectral linear array TDI image sensor.

[0056] Specifically, in some embodiments of the present invention, the super-surface micro-nano structure 203 is in direct contact with the corresponding pixel array 201. That is, there is no other structure, such as a filter, between the super-surface micro-nano structure 203 and the pixel array 201. The super-surface micro-nano structure 203 is directly manufactured on the surface of the pixel array 201 using a semiconductor process, and the received incident light is broadband modulated only according to the super-surface micro-nano structure 203 to obtain modulated light, and the corresponding pixel array 201 receives the modulated light. The production and design costs of precision filters are high, and the hyperspectral linear array TDI image sensor of this solution effectively reduces the cost of the process.

[0057] Specifically, in some embodiments of the present invention, the material of the super-surface micro-nano structure 203 is a light-transmitting material. The material of the super-surface micro-nano structure 203 is a light-transmitting material, which is convenient for receiving incident light and performing broadband modulation on the incident light.

[0058] Specifically, in some embodiments of the present invention, the materials of the multiple super-surface micro-nano structures 203 are the same. The materials of the multiple super-surface micro-nano structures 203 are the same, which makes the process flow more convenient. Specifically, in some embodiments of the present invention, the materials of the super-surface micro-nano structures 203 include but are not limited to organic matter and silicon oxide. In other embodiments, the materials of the super-surface micro-nano structures can also be other materials.

[0059] In other embodiments of the present invention, among the plurality of said super-surface micro-nano structures, at least two of said super-surface micro-nano structures are made of different materials, that is, among the plurality of said super-surface micro-nano structures, at least one of said super-surface micro-nano structures is made of a material different from that of the other said super-surface micro-nano structures.

[0060] The metasurface micro-nano structure 203 performs broadband modulation on the incident light received to obtain modulated light; the corresponding pixel array 201 receives the modulated light. That is, each metasurface micro-nano structure 203 performs broadband modulation on the incident light received by utilizing the different refractive index properties of the incident light due to different shapes, and transmits the modulated light obtained by the broadband modulation to the pixel array 201 corresponding to each metasurface micro-nano structure 203.

[0061] The metasurface micro-nano structure 203 in the metasurface array 202 includes: modulation elements periodically distributed in a plane parallel to the surface of the substrate 200. That is, one metasurface micro-nano structure 203 includes a plurality of modulation elements periodically distributed.

[0062] In some embodiments of the present invention, different metasurface micro-nano structures 203 in the metasurface array 202 have different structures. For example, in different metasurface micro-nano structures 203, at least one of the size, shape, and distribution pattern of the modulators is different.

[0063] Specifically, in some embodiments of the present invention, the cross-sectional shape of the modulation element includes, but is not limited to, a circle, a star, and a circular hole. In other embodiments, the cross-sectional shape of the modulation element also includes other shapes.

[0064] The hyperspectral linear array TDI image sensor further includes a reconstruction module, and the reconstruction module is configured as an acquisition unit and a calculation unit.

[0065] The reconstruction module is suitable for spectrally reconstructing the image to be processed obtained by using the image sensor to obtain spectral information; wherein the acquisition unit is suitable for obtaining the image to be processed based on the received incident light; the operation unit is suitable for using a neural network algorithm to operate on the obtained image to be processed, and obtain spectral information based on the result of the operation.

[0066] The linear array TDI image sensor of the present invention integrates different metasurface micro-nano structures 203 on different pixel arrays 201. These metasurface micro-nano structures 203 perform broadband modulation on the incident spectrum. The TDI linear array sensor obtains multiple groups of images modulated by metasurface micro-nano structures 203 with different structures, and uses a neural network algorithm to calculate and reconstruct a full-spectrum image within a specific wavelength range.

[0067] Correspondingly, an embodiment of the present invention further provides an image processing method, comprising: acquiring an image to be processed using the hyperspectral linear array TDI image sensor as described in any one of the above items; and obtaining spectral information based on the image to be processed.

[0068] The image to be processed is obtained by broadband modulation of the incident light received by the hyperspectral linear array TDI image sensor with the super-surface micro-nano structure 203, thereby obtaining spectral information. The spectral information obtained by the image processing method is not restricted by the number of filters, and full-spectrum image information can be obtained, and the spectral restoration accuracy is higher.

[0069] In the step of obtaining spectral information, spectral information within a preset spectral range is obtained based on the image to be processed.

[0070] In the step of obtaining spectral information, the spectral information is obtained based on the image to be processed and the structure of the metasurface array 202 .

[0071] Specifically, in some embodiments of the present invention, in the step of obtaining spectral information, the spectral information is obtained through a neural network algorithm based on the image to be processed and the structure of the metasurface array 202.

[0072] For examples, please refer to Figure 5 and Figure 6 , Figure 5 Information about a scene; Figure 6 A schematic diagram of a spectral image having a number greater than the number of channels calculated based on image data acquired from each channel for image processing using the hyperspectral linear array TDI image sensor as described in any of the above items.

[0073] The linear array TDI image sensor proposed in the present invention takes a hyperspectral linear array TDI image sensor having a metasurface array (i.e., an 8-channel architecture) with 8 types of metasurface micro-nano structures 203 as an example, and uses the sensor to obtain 8 images to be processed; through the calculation of the neural network algorithm, the full spectrum image information of the images to be processed can be obtained, thereby obtaining all the information of the scene. For example, Figure 6 The data of 16 spectral bands of interest obtained by the neural network algorithm are shown, which is only a part of the capability of the hyperspectral linear array TDI image sensor using a metasurface array (i.e., an 8-channel architecture) with 8 metasurface micro-nano structures 203. The image sensor of the technical solution of the present invention can restore the information of the entire spectrum, so the sensor can output image information of dozens or even hundreds of spectral bands.

[0074] In summary, the super-surface micro-nano structure 203 is located on the pixel array 201, and the super-surface micro-nano structure 203 performs broadband modulation on the incident light received. The super-surface micro-nano structure 203 does not filter out the incident light, so that the brightness of the image of each channel received by the pixel array 201 after broadband modulation by the super-surface micro-nano structure 203 is high, which is beneficial to improve the signal-to-noise ratio of the image; and the hyperspectral linear array TDI image sensor of this scheme does not need to use filters, and the production and design costs of precision filters are high. The hyperspectral linear array TDI image sensor of this scheme effectively reduces the cost of the process; in addition, the hyperspectral linear array TDI image sensor with the super-surface micro-nano structure 203 is used to perform broadband modulation on the incident light received to obtain the image to be processed, thereby obtaining spectral information. The spectral information obtained by the image processing method is not restricted by the number of filters, and full-spectrum image information can be obtained, and the spectral restoration accuracy is higher.

[0075] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A hyperspectral linear array TDI image sensor, characterized in that: include: a substrate, the substrate comprising a pixel region; A plurality of pixel arrays, wherein the plurality of pixel arrays are located on the pixel area, and the plurality of pixel arrays are arranged in sequence along a first direction; A metasurface array, wherein the metasurface array is located on a plurality of the pixel arrays, and the metasurface array comprises a plurality of metasurface micro-nano structures, and the metasurface micro-nano structures correspond one-to-one to the pixel arrays.

2. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: The materials of the multiple super-surface micro-nano structures are the same.

3. The hyperspectral linear array TDI image sensor according to claim 1 or 2, characterized in that: The material of the super surface micro-nano structure includes but is not limited to organic matter and silicon oxide.

4. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: Among the plurality of said super-surface micro-nano structures, at least two of said super-surface micro-nano structures are made of different materials.

5. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: The metasurface micro-nanostructure performs broadband modulation on the received incident light to obtain modulated light; The corresponding pixel array receives the modulated light.

6. The hyperspectral linear array TDI image sensor according to claim 1 or 5, characterized in that: The structure of the metasurface micro-nano structure in the metasurface array is determined based on the spectrum of the incident light and the spectral line band collected by the corresponding pixel array.

7. The hyperspectral linear array TDI image sensor according to claim 5, characterized in that: The metasurface micro-nanostructure in the metasurface array includes: modulation elements periodically distributed in a plane parallel to the substrate surface.

8. The hyperspectral linear array TDI image sensor according to claim 7, characterized in that: The cross-sectional shape of the modulation element includes but is not limited to: circular, star-shaped, and circular hole.

9. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: The super-surface micro-nano structure is in direct contact with the corresponding pixel array.

10. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: On the substrate surface, the projection of the pixel array corresponding to the super-surface micro-nano structure is located within the range of the projection of the super-surface micro-nano structure.

11. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: The number of the pixel arrays is greater than 4.

12. The hyperspectral linear array TDI image sensor according to claim 1, characterized in that: Also includes: A reconstruction module, wherein the reconstruction module is configured as: an acquisition unit and a calculation unit.

13. The hyperspectral linear array TDI image sensor according to claim 12, characterized in that: The reconstruction module is suitable for spectrally reconstructing the image to be processed obtained by using the image sensor to obtain spectral information; wherein the acquisition unit is suitable for obtaining the image to be processed based on the received incident light; the operation unit is suitable for using a neural network algorithm to operate on the obtained image to be processed, and obtain spectral information based on the result of the operation.

14. An image processing method, characterized in that: include: Acquire an image to be processed using the hyperspectral linear array TDI image sensor as described in any one of claims 1 to 13; Based on the image to be processed, spectral information is obtained.

15. The image processing method according to claim 14, characterized in that: In the step of obtaining spectral information, spectral information within a preset spectral range is obtained based on the image to be processed.

16. The image processing method according to claim 15, characterized in that: In the step of obtaining spectral information, the spectral information is obtained based on the image to be processed and the structure of the metasurface array.

17. The image processing method according to claim 16, characterized in that: In the step of obtaining spectral information, the spectral information is obtained through a neural network algorithm based on the image to be processed and the structure of the metasurface array.

Images