A pixel-level discrete integrated snapshot multi-spectral imaging de-cross talk method and system

Abstract

The present application relates to a kind of pixel-level discrete integrated snapshot multispectral imaging method and system of eliminating crosstalk, using the same optical axis arrangement one imaging lens group, filter chip, secondary imaging lens group and imaging chip technology, the way of light path correction by one imaging lens group and secondary imaging lens group is guided light respectively image on filter chip, imaging chip, solve the crosstalk problem between adjacent spectral channels caused by the gap between filter chip and imaging chip;Using m×m filter channel and the pixel-level matching of imaging chip realizes pixel-level spectral imaging;Then through normalization, effective pixel position matrix screening removes invalid pixel, then through the effective pixel position matrix of corresponding working waveband and original spectral image Hadamard product, completely solve the optical crosstalk problem caused by gap and alignment error, reduce the alignment requirement of filter chip and imaging chip in pixel-level discrete integrated snapshot multispectral imaging system of eliminating crosstalk.

Description

Technical Field

[0001] This invention relates to the field of multispectral imaging technology, and more specifically, to a pixel-level discrete integrated snapshot-type multispectral imaging crosstalk reduction method and system. Background Technology

[0002] Spectral imaging technology combines imaging and spectral techniques to simultaneously acquire two-dimensional spatial information from an image and record one-dimensional spectral information corresponding to its physicochemical characteristics. Spectral imaging technology can be divided into hyperspectral imaging and multispectral imaging based on the number of spectral bands acquired. Hyperspectral imaging can acquire hundreds or even thousands of spectral bands, providing rich spectral information; however, this also leads to system complexity, high data storage pressure, and high cost. Multispectral imaging acquires fewer spectral bands, only a few to dozens, making spectral acquisition easier and offering advantages such as smaller data volume, real-time monitoring, compact structure, and lower cost.

[0003] Common methods for implementing multispectral imaging include rotary filters, linearly graded filters, and liquid crystal and acousto-optic tunable filters. While these methods can acquire the required spectral information, they all employ a time-division imaging mode, acquiring information from only one band at a time. This results in long acquisition times and makes them unsuitable for acquiring images in transient scenarios. Using a multispectral filter array can overcome the limitations of time-division imaging. By integrating the multispectral filter array onto the detector, information from multiple bands can be acquired simultaneously.

[0004] There are two main integration methods for multispectral filter arrays (hereinafter referred to as "filter chips"): integrated integration and discrete manufacturing and re-integration. Integrated integration directly constructs the filter array on the sensor surface using micro-nano processes. This method has strong coupling but is complex, has low yield, and requires modifications to the manufacturing process of mature chips, increasing testing costs. In contrast, discrete manufacturing and re-integration manufactures the filter chip and imaging chip separately. This method has a simpler process, higher yield, and can be optimized on their respective mature platforms. It enables standardized decoupling of the filter module and imaging module, facilitating flexible replacement and system expansion. It supports various filtering principles (such as interference films, dyes, nanostructures, etc.) and allows for free design of the number of channels, center wavelength, and bandwidth, without being limited by the manufacturing process of the imaging sensor.

[0005] Currently, many practical pixel-level snapshot multispectral imaging systems still employ a discrete manufacturing and re-integration approach, directly overlaying the filter chip onto the imaging chip. During operation, the incident light beam from the object passes through the filter chip and is captured by the imaging chip, enabling the selection of a specific wavelength band for each pixel. However, in actual integration, the filter chip and imaging chip cannot be perfectly aligned, meaning a gap exists between them. In practice, when the incident light beam enters the filter chip at a non-perpendicular angle, the propagation path within the gap becomes longer, potentially deviating from the target pixel and being captured by neighboring pixels. This results in a single pixel on the imaging chip receiving spectral information from multiple filter channels, causing optical crosstalk and reducing the spectral accuracy of the pixel-level multispectral imaging system. Summary of the Invention

[0006] The technical problem to be solved by this invention is how to eliminate crosstalk in the multispectral imaging process.

[0007] This invention provides a pixel-level discrete integrated snapshot multispectral imaging crosstalk cancellation system, including a filter chip and an imaging chip, as well as a primary imaging lens group and a secondary imaging lens group. The primary imaging lens group, filter chip, secondary imaging lens group, and imaging chip are sequentially arranged on the same optical axis. The primary imaging lens group is an image-side telecentric optical path, and the coated surface of the filter chip is located at the image plane of the primary imaging lens group. The filter chip includes... The primary imaging lens group has a filter channel; the secondary imaging lens group is a telecentric optical path on the object side, the object plane of the secondary imaging lens group is coplanar with the coating surface of the filter chip, and the imaging chip is located at the image plane of the secondary imaging lens group; the primary imaging lens group is also provided with a focusing structure for adjusting the object distance so that the image plane coincides with the coating surface of the filter chip.

[0008] Compared with existing technologies, the system of this application has the following advantages: This application adopts a primary imaging mirror group, a filter chip, a secondary imaging mirror group, and an imaging chip technology arranged on the same optical axis. The primary imaging mirror group guides the light to image onto the surface of the filter chip through optical path correction. Then, the secondary imaging mirror group corrects the optical path of the light processed by the filter chip and images it onto the imaging chip. This solves the crosstalk problem between adjacent spectral channels caused by the gap between the filter chip and the imaging chip; simultaneously, it employs... Pixel-level spectral imaging is achieved through pixel-level matching of multiple filtering channels with the imaging chip.

[0009] In one possible implementation, the filter chip is a multispectral filter array, and the filter chip has... Spectral information for each working band; each filter channel includes A subpixel array consisting of imaging pixels.

[0010] Compared with existing technologies, based on Each filter channel includes [number] filter channels. The design of a subpixel array composed of imaging pixels provides flexibility in spectral imaging. By adjusting the number of filter channels and working bands, it can adapt to different application scenarios. Furthermore, the one-to-many matching of filter channels and pixels greatly improves the utilization rate of light energy and adapts to low-light environments.

[0011] In one possible implementation, the imaging chip is a monochrome CMOS sensor.

[0012] Compared to existing technologies, monochrome CMOS sensors can achieve more accurate spectral responses than traditional color sensors.

[0013] A pixel-level discrete integrated snapshot-type multispectral imaging crosstalk reduction method includes:

[0014] Step 1: Build the pixel-level discrete integrated snapshot multispectral imaging anti-crosstalk system described above, set the imaging target in front of the pixel-level discrete integrated snapshot multispectral imaging anti-crosstalk system, and adjust the object distance of the primary imaging lens group.

[0015] Step 2, set up the imaging lens group in sequence. Narrowband filters in each working band were photographed to obtain... The original spectral images of each working band;

[0016] Step 3: Normalize the original spectral image, and select the pixel with the highest grayscale value that is completely covered by the corresponding filter channel in each sub-pixel array to generate... The effective pixel location matrix of each working band;

[0017] Step 4: Perform a Hadamard product between the effective pixel location matrix of the corresponding working band and the original spectral image to obtain the processed image. Spectral images of each working band;

[0018] Step 5, process the... The spectral images of each working band are de-mosaiced to output a crosstalk-free multispectral image.

[0019] Compared with existing technologies, the method of this application has the following advantages: This application is based on a pixel-level discrete integrated snapshot multispectral imaging crosstalk reduction system to acquire... The original spectral images of each working band are normalized and invalid pixels are removed by filtering with an effective pixel position matrix. Then, the effective pixel position matrix of the corresponding working band is used to perform a Hadamard product with the original spectral image to completely solve the optical crosstalk problem caused by gaps and alignment errors, and reduce the alignment requirements of the filter chip and the imaging chip in the pixel-level discrete integrated snapshot multispectral imaging crosstalk cancellation system.

[0020] In one possible implementation, step 1, which involves building a pixel-level discrete integrated snapshot multispectral imaging crosstalk reduction system, specifically includes:

[0021] The primary imaging lens group, the filter chip, and the secondary imaging lens group are housed in the same imaging lens tube, which is then matched and connected to a monochrome industrial camera. The object distance of the primary imaging lens group is adjusted by a focusing structure to ensure that the object surface of the target is clearly imaged on the filter chip, while the secondary imaging lens group converges the filtered image onto the imaging chip.

[0022] In one possible implementation, adjusting the object distance of the primary imaging lens group in step 1 specifically includes:

[0023] Calculate the optimal object distance for imaging targets The calculation formula is:

[0024] ;

[0025] In the formula, Indicates the focal length of the single-image lens group. This indicates the distance between the filter chip and the image plane of the primary imaging lens group.

[0026] In one possible implementation, in step 2 The narrowband filters for each operating band represent , captured by the camera The set of original spectral images for each working band is represented as: , Represented as the first The original spectral images of each working band.

[0027] In one possible implementation, step 3 specifically includes:

[0028] Step 301, the filter chip is made of It consists of several filter channels, covering Each operating band, each filter channel consists of... It consists of individual pixels; the resolution of each original spectral image is consistent with the resolution of the corresponding working band, defining the working band. The coverage area is:

[0029] ;

[0030] In the formula, Represents the pixel coordinates of the original spectral image. ;

[0031] Step 302, from the original spectral image Cut out the area to be covered area The size is , Indicates the total number of rows in the filter channel. Indicates the total number of columns in the filter channels;

[0032] Step 303, Calculate the region Normalized image:

[0033] ;

[0034] In the formula, Indicates the region The coordinates of the inner pixel;

[0035] Step 304, based on the maximum grayscale value Filter the pixels in the sub-pixel array to obtain Effective pixel location matrix of spectral images for each working band The filter condition expression is:

[0036] .

[0037] Compared with existing technologies, the normalization process eliminates interference from ambient light, ensuring that the selected pixels are fully covered by the filtering channel. The larger the sub-pixel array, the more effective pixels there are, thus improving the light energy utilization rate.

[0038] In one possible implementation, step 4 involves dividing the effective pixel position matrix of the corresponding working band. The Hadamard product with the spectral image is expressed as:

[0039] ;

[0040] In the formula, Represents the Hadamard product. Indicates the first The spectral image after Hadamard integration to remove crosstalk in each working band.

[0041] Compared with existing technologies, the Hadamard energy integration method directly filters out crosstalk light intensity from non-corresponding filter channels, improves spectral purity, and has a simple overall operation, which is beneficial for embedded real-time processing.

[0042] In one possible implementation, step 5 involves de-mosaicing the processed spectral images of different working bands, including:

[0043] Spectral images for each working band The expression for bilinear interpolation de-mosaic processing is:

[0044] ;

[0045] In the formula, Represents distance weights. This represents the effective pixel neighborhood.

[0046] Compared with existing technologies, bilinear interpolation demosaicing is used to reconstruct pixels in the missing area by interpolating effective pixels in the neighborhood, thus mitigating the resolution loss caused by the one-to-many matching of filter channels and pixels. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of a specific embodiment 1 of the present invention;

[0048] Figure 2 This is a grayscale image of the filter channel and the imaging pixels matched in 3×3 in specific embodiment 2 of the present invention;

[0049] Figure 3 This is a grayscale image of the filter channel and the imaging pixels matched 2×2 in specific embodiment 2 of the present invention;

[0050] Figure 4 This is a diagram showing the relationship between the effective pixel ratio caused by misalignment between the filter chip and the imaging chip in specific embodiment 2 of the present invention.

[0051] Figure 5 This is a grayscale image showing the matching of the filter channel and the 10×10 imaging pixels in specific embodiment 2 of the present invention.

[0052] Explanation of reference numerals in the attached figures:

[0053] 1. Primary imaging lens group; 2. Filter chip; 3. Secondary imaging lens group; 4. Imaging chip. Detailed Implementation

[0054] First, those skilled in the art should understand that these embodiments are merely used to explain the technical principles of the embodiments of this application and are not intended to limit the scope of protection of the embodiments of this application. Those skilled in the art can make adjustments as needed to adapt to specific application scenarios.

[0055] In the description of the embodiments of this application, it should be noted that, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application based on the specific circumstances.

[0056] In the embodiments of this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0057] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments. Specific Implementation Example 1

[0059] See Figure 1 As shown in the figure, this application discloses a pixel-level discrete integrated snapshot multispectral imaging anti-crosstalk system, including a filter chip 2, an imaging chip 4, a primary imaging lens group 1 and a secondary imaging lens group 3, wherein the primary imaging lens group 1, the filter chip 2, the secondary imaging lens group 3 and the imaging chip 4 are sequentially arranged on the same optical axis.

[0060] The primary imaging lens group 1 is an image-side telecentric optical path that images the object plane of the target onto the image plane; the primary imaging lens group 1 is also provided with a focusing structure for adjusting the object distance so that the image plane coincides with the coating surface of the filter chip 2.

[0061] The coating surface of the filter chip 2 is located at the image plane of the primary imaging lens group 1; the filter chip 2 includes... Each filter channel includes [number] filter channels; each filter channel includes [number] filter channels. The sub-pixel array consists of imaging pixels, and the filter chip 2 is a multispectral filter array, including... Spectral information for each working band; embodiments of this application It adopts a four-spectral filtering channel design, with each filtering channel measuring 7.5μm×7.5μm. The four filtering channels select 500nm, 550nm, 600nm, and 650nm as the filtering bands, respectively.

[0062] The secondary imaging mirror group 3 is a telecentric optical path on the object side, and the object plane of the secondary imaging mirror group 3 is coplanar with the coating surface of the filter chip 2.

[0063] The imaging chip 4 is a monochrome CMOS sensor. In this embodiment, the pixel size of the imaging chip 4 is 2.5μm×2.5μm, which satisfies the requirement that one filter channel matches a sub-pixel array composed of 3×3 imaging pixels. The imaging chip 4 is disposed at the image plane of the secondary imaging lens group 3. Specific Implementation Example 2

[0065] The working principle of this application is as follows:

[0066] When illuminated using the operating wavelength of a specific filter channel, the effective pixels in the sub-pixel array completely covered by the corresponding filter channel have higher grayscale values ​​and experience no crosstalk from other channels during imaging. Pixels not completely covered by the corresponding filter channel have lower grayscale values ​​and are more likely to collect light from multiple channels during imaging, resulting in crosstalk. Based on this principle, this application obtains the effective pixel positions for each filter channel by selecting the effective pixels corresponding to the filter channel and sequentially calibrating all filter channels. The effective pixel positions for each channel are recorded and saved as an effective pixel position matrix (elements at effective pixel positions are 1, and elements at other positions are 0). Subsequently, this matrix is ​​multiplied by the captured image using the Hadamard product to obtain the processed image.

[0067] This application's embodiments address crosstalk caused by gap or alignment accuracy issues between the filter chip 2 and the imaging chip 4, such as... Figure 2 As shown, a single filter channel corresponds to a sub-pixel array consisting of 3×3 imaging pixels, and the fully aligned state is as follows. Figure 2 As shown in (a), when the filter chip 2 shifts or rotates, some imaging pixels in the sub-pixel array are not completely covered by the corresponding filter channel, as shown in (a). Figure 2 As shown in (b) of the diagram.

[0068] Based on the above problems, this application discloses a pixel-level discrete integrated snapshot multispectral imaging crosstalk reduction method according to embodiment 1, including:

[0069] Step 1: Construct a pixel-level discrete integrated snapshot multispectral imaging crosstalk cancellation system. Set the imaging target in front of the pixel-level discrete integrated snapshot multispectral imaging crosstalk cancellation system and adjust the object distance of the primary imaging lens group 1; specifically including:

[0070] The primary imaging lens group 1, the filter chip 2, and the secondary imaging lens group 3 are set in the same imaging lens barrel. The imaging lens barrel is threadedly connected to a monochrome industrial camera based on the imaging chip 4. The imaging chip 4 on the monochrome industrial camera is located at the image plane of the secondary imaging lens group 3.

[0071] The imaging target is placed in front of the pixel-level discrete integrated snapshot multispectral imaging anti-crosstalk system, and the object distance is adjusted by the focusing structure; the imaging target is a uniform LED panel light source or a diffuse reflector.

[0072] The object distance of the primary imaging lens group 1 is adjusted by the focusing structure so that the object plane of the imaging target is clearly imaged on the surface of the filter chip 2, and the secondary imaging lens group 3 converges the filtered image onto the imaging chip 4.

[0073] Regarding the process of adjusting the object distance, the embodiments of this application employ the calculation of the optimal object distance of the imaging target. The calculation formula is:

[0074] ;

[0075] In the formula, This indicates the focal length of the single-image lens group 1. This indicates the distance between filter chip 2 and the image plane of primary imaging lens group 1.

[0076] Step 2, set up the following in sequence in front of the primary imaging lens group 1 Narrowband filters in each working band were photographed to obtain... The original spectral images of each working band; in this embodiment, the working bands are respectively placed in front of the imaging lens group 1. , , , After applying a narrowband filter, the images were taken to obtain the original spectral images of the four working bands.

[0077] Step 3: Normalize the original spectral image, and select the pixel with the highest grayscale value that is completely covered by the corresponding filter channel in each sub-pixel array to generate... The effective pixel location matrix for each working band; specifically including:

[0078] Step 301, the filter chip 2 is composed of It consists of several filter channels, covering Each operating band, each filter channel consists of... It consists of individual pixels; the resolution of each original spectral image is consistent with the resolution of the corresponding working band, defining the working band. The coverage area is:

[0079] ;

[0080] In the formula, Represents the pixel coordinates of the original spectral image. ;

[0081] Step 302, from the original spectral image Cut out the area to be covered area The size is , Indicates the total number of rows in the filter channel. Indicates the total number of columns in the filter channels;

[0082] Step 303, Calculate the region Normalized image:

[0083] ;

[0084] In the formula, Indicates the region The coordinates of the inner pixels, after image normalization, are as follows: Figure 2 As shown in (b) above, each imaging pixel has a corresponding grayscale value;

[0085] Step 304, based on the maximum grayscale value Filter the pixels in the sub-pixel array to obtain Effective pixel location matrix of spectral images for each working band The filter condition expression is:

[0086] .

[0087] The effective pixel position matrices obtained in the embodiments of this application are as follows: .

[0088] Step 4: Obtain the effective pixel location matrix for the corresponding working band. Perform a Hadamard product with the original spectral image to obtain the processed image. Spectral images of each working band; the expression is:

[0089] ;

[0090] In the formula, Represents the Hadamard product. Indicates the first The spectral image after Hadamard integration to remove crosstalk in each working band.

[0091] Step 5, process the... The spectral images of each working band undergo mosaic removal processing to output a crosstalk-free multispectral image; including:

[0092] Spectral images for each working band The expression for bilinear interpolation de-mosaic processing is:

[0093] ;

[0094] In the formula, Represents distance weights. This represents the effective pixel neighborhood.

[0095] Furthermore, a single filter channel corresponding to a 2×2 sub-pixel array can typically only select one pixel as a valid pixel; see [reference needed]. Figure 3 The filter chip 2 and the imaging chip 4 are fully aligned, such as Figure 3 As shown in (a), the 16 pixels are divided into four 2×2 sub-pixel arrays, each completely covered by four filtering channels, and all 16 pixels are valid pixels. Figure 3 In the diagram, (b)-(d) represent a non-perfectly aligned state, but it can be observed that there are still valid pixels in each sub-pixel array. Figure 3 In (b), the filter chip 2 and the imaging chip 4 are horizontally offset, and two pixels in each sub-pixel array are not completely covered by the corresponding filter channel, while the remaining two pixels are valid pixels; Figure 3 In (c), the filter chip 2 and the imaging chip 4 are offset in an oblique direction. In each sub-pixel array, 3 pixels are not completely covered by the corresponding filter channel, and the remaining 1 pixel is the effective pixel. Figure 3 In (d), a rotational offset occurs between filter chip 2 and imaging chip 4. In each sub-pixel array, 3 pixels are not completely covered by the corresponding filter channel, and the remaining 1 pixel is a valid pixel. Regardless of the alignment error produced by filter chip 2, there will always be at least 1 valid pixel in each sub-pixel array. In actual use, a large amount of light energy is wasted, resulting in low light energy utilization. Therefore, a scheme where a single filter channel corresponds to a larger sub-pixel array can be adopted. Figure 4 This demonstrates the proportion of the minimum number of effective pixels in a sub-pixel array to its total number of pixels as the size of the sub-pixel array gradually increases when a single filter channel covers a single n×n sub-pixel array. Figure 2 and Figure 5 Taking 3×3 and 10×10 sub-pixel arrays corresponding to a single filter channel as examples, respectively, Figure 2 (a) and Figure 5 (a) shows the state where the filter chip 2 and the imaging chip 4 are fully aligned; Figure 2 (b) and Figure 5 In (b), a rotational offset occurs between the filter chip 2 and the imaging chip 4. Under the same deviation conditions, each filter channel can completely correspond to multiple pixels, greatly improving the light energy utilization rate and enabling the scheme to handle complex situations such as low light. The method of this application can reduce the alignment accuracy between the imaging chip 4 and the filter chip. In the case of incomplete alignment, effective pixels are selected to participate in subsequent imaging work, avoiding spectral crosstalk.

[0096] In the description of the embodiments of this application, it should be noted that the terms "inner" and "outer" and other terms indicating direction or positional relationship are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or component must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this application.

[0097] In the description of this application, the references to terms such as "an embodiment," "some embodiments," "in this embodiment," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A pixel-level discrete integrated snapshot multi-spectral imaging de-cross talk system comprising a filtering chip (2) and an imaging chip (4), characterized in that, Also include a primary imaging lens group (1) and secondary imaging lens group (3), the primary imaging lens group (1), filter chip (2), secondary imaging lens group (3) and imaging chip (4) are sequentially arranged on the same optical axis, the primary imaging lens group (1) is image side telecentric optical path, the filter chip (2) is provided at the image surface of the primary imaging lens group (1) The film surface of;The filter chip (2) includes A filter channel;The secondary imaging lens group (3) is object side telecentric optical path, the object surface of the secondary imaging lens group (3) and the film surface of filter chip (2) are coplanar, the imaging chip (4) is provided at the image surface of secondary imaging lens group (3);The primary imaging lens group (1) is also provided with the focusing structure for adjusting the object distance to make the image surface coincide with the film surface of filter chip (2) The filter chip (2) is a multi-spectrum filter array, and the filter chip (2) has spectrum information of K working wavebands; each filter channel includes a sub-pixel array composed of M* N imaging pixels. A sub-pixel array composed of M* N imaging pixels.

2. The pixel-level discrete integrated snapshot multispectral imaging de-cross talk system according to claim 1, wherein, The imaging chip (4) is a monochrome CMOS sensor.

3. A pixel-level discrete integrated snapshot multi-spectral imaging de-cross talk method, characterized in that, Comprise: Step 1, set up the pixel-level discrete integrated snapshot multispectral imaging system of claim 2, set the imaging target in front of the pixel-level discrete integrated snapshot multispectral imaging system, and adjust the object distance of the primary imaging lens group (1); Step 2, set a narrow-band filter of one working waveband in front of the primary imaging lens group (1) and take a picture to obtain an original spectral image of the working waveband Step 3, set a narrow-band filter of another working waveband in front of the primary imaging lens group (1) and take a picture to obtain an original spectral image of the working waveband Step 4, set a narrow-band filter of another working waveband in front of the primary imaging lens group (1) and take a Step 3, normalizing the original spectral image, screening the pixel with the highest gray value in each sub-pixel array which is completely covered by the corresponding filter channel, and generating an effective pixel position matrix of the working waveband; Step 4, Hadamard product of the effective pixel position matrix of the corresponding working waveband and the original spectral image is obtained to obtain the processed spectral image of the working waveband Step 4, Hadamard product of the effective pixel position matrix of the corresponding working waveband and the original spectral image is obtained to obtain the processed spectral image of the working waveband Step 5, the processed spectral images of the working waveband are demosaicked to output the desmeared multispectral images.

4. The pixel-level discrete integrated snapshot multispectral imaging de-cross talk method according to claim 3, characterized in that, The step 1 of setting up the pixel-level discrete integrated snapshot multispectral imaging system specifically comprises: The primary imaging lens group (1), the filter chip (2) and the secondary imaging lens group (3) are arranged in the same imaging lens barrel, the imaging lens barrel is matched and connected with the monochrome industrial camera; the object distance of the primary imaging lens group (1) is adjusted through the focusing structure, so that the object plane of the imaging target is clearly imaged on the filter chip (2), and the filtered image is converged to the imaging chip (4) by the secondary imaging lens group (3).

5. The pixel-level discrete integrated snapshot multispectral imaging system of claim 3, wherein the step 1 of adjusting the object distance of the primary imaging lens group (1) specifically comprises: The step 3 specifically comprises: Computing an optimal object distance for an imaging target The formula is: ； wherein denotes the focal length of the primary imaging lens group (1), denotes the distance of the filter chip (2) from the image plane of the primary imaging lens group (1).

6. The pixel-level discrete integrated snapshot multispectral imaging de-cross talk method according to claim 3, wherein, In step 2 The narrowband filters for each operating band represent , captured by the camera The set of original spectral images for each working band is represented as: , Represented as the first The original spectral images of each working band.

7. The pixel-level discrete integrated snapshot multispectral imaging de-cross talk method according to claim 6, characterized in that, The step 5 of de-mosaicking the processed spectral images of different working wavebands comprises: Step 301, the filter chip (2) is composed of It consists of several filter channels, covering Each operating band, each filter channel consists of... It consists of individual pixels; the resolution of each original spectral image is consistent with the resolution of the corresponding working band, defining the working band. The coverage area is: ； In the formula, denotes the pixel coordinates of the original spectral image, ; Step 302, cropping from the original spectral image the region belonging to the coverage area , the size of which is , represents the total number of rows of filter channels, represents the total number of columns of filter channels;​ Step 303, calculating the region of the normalized image: ； In the formula, indicates the region coordinates of the inner pixel; Step 304, based on the maximum gray value Screening the pixels in the sub-pixel array to obtain The effective pixel position matrix of the spectral image of the working waveband The screening condition expression is: 。 8. The pixel-level discrete integrated snapshot multispectral imaging de-cross talk method according to claim 7, characterized in that, The step 4 will make the effective pixel position matrix of the corresponding working waveband Hadamard product with the original spectral image, the expression is: ； wherein represents the Hadamard product, represents the Hadamard product, represents the Hadamard product of the spectral image of the first working wavelength band.

9. The pixel-level discrete integrated snapshot multispectral imaging de-cross talk method according to claim 8, characterized in that, ​ spectral images for each working wavelength band A bilinear interpolation demosaicing process is performed, expressed as: ； wherein denotes the distance weight, denotes the neighborhood of valid pixels.

Images