A single-pixel hyperspectral imaging system and method based on dual filters

Abstract

The present invention discloses a single-pixel hyperspectral imaging system and method based on dual filters, belonging to the field of hyperspectral imaging. The imaging system comprises a light source, a digital micromirror device, a first lens, a second lens, a third lens, a first flat beam splitter, a second flat beam splitter, a Cos filter, a Sin filter, a first single-pixel detector, a second single-pixel detector and a third single-pixel detector. The provided imaging method obtains the imaging of a hyperspectral object through a data acquisition stage, an image reconstruction stage, a material spectrum analysis stage and a hyperspectral imaging stage, effectively reducing the current hyperspectral imaging cost and the current hyperspectral imaging calculation amount, and has broad application prospects.

Description

Technical Field

[0001] The present invention belongs to the field of hyperspectral imaging, and in particular relates to a single-pixel hyperspectral imaging system and method based on dual filters. Background Art

[0002] Hyperspectral imaging is a technology that combines traditional imaging and spectral detection. It can obtain information about the spatial and spectral dimensions of objects, and has high spatial resolution and spectral resolution. For objects with highly overlapping spectral information, hyperspectral imaging can accurately analyze the characteristics of the object to be measured by combining spectral information. The current mainstream hyperspectral imaging technology mainly includes three methods: spatial scanning, spectral scanning, and snapshot. Spatial scanning is to scan and detect the object to be measured point by point to obtain the spectral information of each point. Spectral scanning is to add a variety of narrow-band filters in front of the camera to obtain images through these filters respectively. Snapshot is to obtain the hyperspectral information of the object through light field encoding and complex post-processing algorithm processing and perform post-analysis. Therefore, hyperspectral imaging technology has been widely used in agriculture, remote sensing, geology, medical diagnosis and other fields. However, despite these advantages of this technology and the fact that there are many commercial products on the market, most researchers do not use it. This is because when using traditional linear decomposition or principal component analysis to post-process multidimensional hyperspectral data, they face problems such as large data calculation volume and high cost of hyperspectral imaging equipment. Moreover, traditional hyperspectral cameras rely on array detectors, which may be difficult for spectral analysis in non-visible light bands, because the array detectors in these bands are difficult to process and the equipment is expensive, making it difficult to achieve full popularization. Considering practical applications and scientific research needs, it is necessary to find a method that can reduce the cost of hyperspectral imaging and reduce the amount of current hyperspectral imaging calculations. In summary, existing hyperspectral imaging still has problems such as high hardware equipment cost, large data volume, and complex post-data processing. Summary of the invention

[0003] The purpose of the present invention is to provide a single-pixel hyperspectral imaging system and method based on dual filters to solve the problems of high hardware equipment cost, large data volume and complex post-data processing in the existing hyperspectral imaging technology.

[0004] To achieve the above-mentioned object, the present invention provides a single-pixel hyperspectral imaging system based on dual filters, comprising a light source, a digital micromirror device, a first lens, a second lens, a third lens, a first flat beam splitter, a second flat beam splitter, a Cos filter, a Sin filter, a first single-pixel detector, a second single-pixel detector and a third single-pixel detector;

[0005] Light source: emits steady white light;

[0006] Digital micromirror device: generates random coding patterns for optical coding of the object to be measured;

[0007] First lens: projects the random coded pattern on the digital micromirror device onto the object to be tested;

[0008] Second lens: collects the total light intensity after encoding the object to be measured;

[0009] The third lens: turns the light beam into quasi-collimated light;

[0010] The first flat beam splitter: splits the light into two beams, including reflected light and transmitted light, where the energy ratio of reflected light to transmitted light is 3:7;

[0011] Second flat beam splitter: splits the transmitted light after passing through the first flat beam splitter into reflected light and transmitted light again, where the ratio of reflected light energy to lens light energy is 1:1

[0012] Cos filter: filters the reflected light after the second flat beam splitter;

[0013] Sin filter: filters the transmitted light passing through the second flat beam splitter;

[0014] The first single pixel detector collects the total light intensity after being reflected by the first flat beam splitter;

[0015] The second single-pixel detector collects the total light intensity after being filtered by the Cos filter;

[0016] The third single-pixel detector collects the total light intensity after being filtered by the Sin filter.

[0017] The present invention also provides a single-pixel hyperspectral imaging method based on dual filters, comprising the following steps:

[0018] Step 1, data collection stage: after the light source is irradiated on the digital micromirror device and passes through the built system, the total light intensity of the object to be measured without passing through the filter and the total light intensity after passing through the Cos filter and the Sin filter are obtained respectively;

[0019] Step 2, image reconstruction stage: Use the compressed sensing algorithm to reconstruct the image of the object, and reconstruct the image of the object to be tested without the filter according to the total light intensity obtained in step 1. all And the images O after passing through the Cos filter and Sin filter respectively cos and image O sin ;

[0020] Step 3, material spectrum analysis stage: calculate the phasor point coordinates corresponding to each pixel point in the reconstructed image in step 2, and obtain the phasor spectrum distribution diagram of the image. The specific expression is as follows:

[0021]

[0022] In the formula, ( x0,y 0) To reconstruct the spatial position coordinates of any pixel point in the image,

[0023] G ( x0,y 0) Represents the G coordinate value at the position (x0, y0), S ( x0,y 0) Represents the S coordinate value at the position (x0, y0), O ( x0,y 0)cos Represents image O cos The pixel value at position (x0, y0) in the image, O ( x0,y 0)all Represents image O all The pixel value at position (x0, y0) in the image, O ( x0,y 0)sin Represents image O sin The pixel value at position (x0, y0) in the image;

[0024] Step 4, reconstruction of the hyperspectral color image stage; according to the phasor spectrum distribution diagram obtained in step 3, the coordinates of each position point in the phasor spectrum diagram represent a characteristic spectrum, so that the characteristic spectrum corresponding to each pixel in the image can be obtained, and the color corresponding to the pixel is reconstructed according to the characteristic spectrum. The above operation is performed on each pixel in the reconstructed image to obtain a color image of the hyperspectral object.

[0025] Preferably, the specific process of step 1 is as follows:

[0026] S11, irradiating a light source on a digital micromirror device, generating a coding pattern using the digital micromirror device, optically encoding the object to be measured, and projecting the coding pattern onto the object to be measured through a first lens;

[0027] S12, collecting the total light intensity from the encoded object through the second lens and the third lens;

[0028] S13, using a first flat beam splitter and a second flat beam splitter to split the encoded total light intensity into three beams;

[0029] S14, using a Cos filter and a Sin filter to filter the reflected light and the transmitted light passing through the second flat beam splitter respectively;

[0030] S15. Build a single-pixel computational imaging system, use the first single-pixel detector to collect the total intensity of light after reflection by the first flat beam splitter, use the second single-pixel detector to collect the total intensity of light after filtering by the Cos filter, and use the third single-pixel detector to collect the total intensity of light after filtering by the Sin filter.

[0031] Preferably, in S13, the encoded total light intensity is divided into three beams using the first flat beam splitter and the second flat beam splitter, specifically: the collected light is divided into two beams by the first flat beam splitter, the intensity value of one beam is collected by the first single-pixel detector, and the other beam is divided into two beams again by the second flat beam splitter, and respectively collected by the second single-pixel detector and the third single-pixel detector.

[0032] Preferably, the single-pixel computational imaging system is constructed by one or more of a differential single-pixel imaging algorithm, a compressed sensing algorithm, an iterative or non-iterative algorithm, a linear or nonlinear algorithm, and a deep learning algorithm.

[0033] Therefore, the present invention adopts the above-mentioned single-pixel hyperspectral imaging system and method based on dual filters, based on spectral phasor analysis, and uses a single-pixel detector to achieve material spectral resolution and hyperspectral imaging of the reconstructed image, effectively reducing the current cost of hyperspectral imaging, reducing the current amount of computational post-processing of hyperspectral data, and has high application value and broad application prospects.

[0034] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a structural schematic diagram of a single-pixel hyperspectral imaging system based on dual filters of the present invention;

[0036] Figure 2 : is a transmission spectrum diagram of the Cos filter in an embodiment of the present invention;

[0037] Figure 3 is a transmission spectrum diagram of the Sin filter in an embodiment of the present invention;

[0038] Figure 4 Schematic diagram of phasor spectrum calibration points established by using Cos filter and Sin filter in an embodiment of the present invention;

[0039] Figure 5 This is an application example result of detecting an actual object in an embodiment of the present invention;

[0040] Among them, 1. light source, 2. digital micromirror device, 3. first lens, 4. object to be measured, 5. second lens, 6. third lens, 7. first flat beam splitter, 8. second flat beam splitter, 9. Sin filter, 10. Cos filter, 11. first single-pixel detector, 12. second single-pixel detector, 13. third single-pixel detector. DETAILED DESCRIPTION

[0041] The following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] See also Figure 1-5 , a single-pixel hyperspectral imaging system based on dual filters, comprising a light source 1, a digital micromirror device 2, a first lens 3, a second lens 5, a third lens 6, a first flat beam splitter 7, a second flat beam splitter 8, a Cos filter 10, a Sin filter 9, a first single-pixel detector 11, a second single-pixel detector 12 and a third single-pixel detector 13;

[0043] Light source 1: emits steady white light;

[0044] Digital micromirror device 2: generates a random coding pattern for optical coding of the object to be tested, wherein the random pattern is a pattern constructed based on a Fourier basis or a pattern optimized based on a Hadamard matrix, etc.;

[0045] First lens 3: projects the random coding pattern on the digital micromirror device onto the object to be measured;

[0046] The second lens 5 is used to collect the total light intensity after encoding the object 4 to be measured;

[0047] The third lens 6: converts the light beam into quasi-collimated light;

[0048] The first flat beam splitter 7: splits the light into two beams, including reflected light and transmitted light, wherein the energy ratio of the reflected light energy to the transmitted light energy is 3:7;

[0049] The second flat beam splitter 8: splits the transmitted light after passing through the first flat beam splitter into reflected light and transmitted light again, where the ratio of reflected light energy to lens light energy is 1:1

[0050] Cos filter 10: filters the reflected light passing through the second flat beam splitter;

[0051] Sin filter 9: filters the transmitted light passing through the second flat beam splitter;

[0052] The first single pixel detector 11 collects the total light intensity after being reflected by the first flat beam splitter;

[0053] The second single pixel detector 12 collects the total light intensity after being filtered by the Cos filter;

[0054] The third single pixel detector 13 collects the total light intensity after being filtered by the Sin filter.

[0055] A single-pixel hyperspectral imaging method based on dual filters comprises the following steps:

[0056] Step 1, data collection stage: After the light source is irradiated on the digital micromirror device, the total light intensity of the object to be measured without passing through the filter and the total light intensity after passing through the Cos filter and the Sin filter are obtained through the built system; the specific process is as follows:

[0057] S11, irradiating a light source on a digital micromirror device, generating a coding pattern using the digital micromirror device, optically encoding the object to be measured, and projecting the coding pattern onto the object to be measured through a first lens;

[0058] S12, collecting the total light intensity from the encoded object through the second lens and the third lens;

[0059] S13, using a first flat beam splitter and a second flat beam splitter to divide the encoded total light intensity into three beams of light; specifically: using the first flat beam splitter to divide the collected light into two beams, wherein the intensity value of one beam is collected by the first single-pixel detector, and the other beam of light is again divided into two beams by the second flat beam splitter, and respectively collected by the second single-pixel detector and the third single-pixel detector;

[0060] S14, using a Cos filter and a Sin filter to filter the reflected light and the transmitted light passing through the second flat beam splitter respectively;

[0061] S15. Build a single-pixel computational imaging system, use a first single-pixel detector to collect the total light intensity after being reflected by the first flat beam splitter, use a second single-pixel detector to collect the total light intensity after being filtered by a Cos filter, and use a third single-pixel detector to collect the total light intensity after being filtered by a Sin filter; wherein the single-pixel computational imaging system is built by one or more of a differential single-pixel imaging algorithm, a compressed sensing algorithm, an iterative or non-iterative algorithm, a linear or nonlinear algorithm, and a deep learning algorithm;

[0062] Step 2, image reconstruction stage: Use the compressed sensing algorithm to reconstruct the image of the object, and reconstruct the image of the object to be tested without the filter according to the total light intensity obtained in step 1. all And the images O after passing through the Cos filter and Sin filter respectivelycos and image O sin ;

[0063] Step 3, material spectrum analysis stage: calculate the phasor point coordinates corresponding to each pixel point in the reconstructed image in step 2, and obtain the phasor spectrum distribution diagram of the image. The specific expression is as follows:

[0064]

[0065] In the formula, ( x0,y 0) To reconstruct the spatial position coordinates of any pixel point in the image,

[0066] G ( x0,y 0) Represents the G coordinate value at the position (x0, y0), S ( x0,y 0) Represents the S coordinate value at the position (x0, y0), O ( x0,y 0)cos Represents image O cos The pixel value at position (x0, y0) in the image, O ( x0,y 0)all Represents image O all The pixel value at position (x0, y0) in the image, O ( x0,y 0)sin Represents image O sin The pixel value at position (x0, y0) in the image;

[0067] Step 4, hyperspectral color image reconstruction stage; according to the phasor spectrum distribution diagram obtained in step 3, each position point coordinate in the phasor spectrum diagram represents a characteristic spectrum, so that the characteristic spectrum corresponding to each pixel in the image can be obtained, and the color corresponding to the pixel is reconstructed according to the characteristic spectrum. The above operation is performed on each pixel in the reconstructed image to obtain a color image of the hyperspectral object, such as Figure 4 As shown, a phase spectrum distribution diagram established using a Cos filter and a Sin filter is shown, wherein each point in the diagram represents a type of hyperspectral data, which can be represented by the same characteristic spectrum, specifically a characteristic spectrum with a central wavelength of 500nm and a broadening parameter of 50nm.

[0068] Therefore, the present invention adopts the above-mentioned single-pixel hyperspectral imaging system and method based on dual filters, and uses a digital micromirror device DMD to perform optical encoding on the object to be measured. The total light intensity after encoding is collected by the second lens and the third lens and passes through two flat beam splitters in sequence. The total light intensity is divided into three beams of light with equal energy by the flat beam splitters, two of which pass through the Cos filter and the Sin filter respectively. The single-pixel detector collects the total light intensity values ​​of the three beams of light respectively, and uses a single-pixel imaging reconstruction method to reconstruct three images of the object to be measured without filtering and after passing through the Cos and Sin filters respectively. Combined with the spectral phasor analysis method, the phasor point coordinates of the object spectrum are obtained, and the hyperspectral analysis and imaging of the object to be measured are finally realized according to the position of the phasor point coordinates, which effectively reduces the current cost of hyperspectral imaging and reduces the amount of calculation for post-processing of hyperspectral data. It has high application value and broad application prospects.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A single-pixel hyperspectral imaging system based on dual filters, characterized in that: It includes a light source, a digital micromirror device, a first lens, a second lens, a third lens, a first flat beam splitter, a second flat beam splitter, a Cos filter, a Sin filter, a first single-pixel detector, a second single-pixel detector and a third single-pixel detector; Light source: emits steady white light; Digital micromirror device: generates random coding patterns for optical coding of the object to be measured; First lens: projects the random coded pattern on the digital micromirror device onto the object to be tested; Second lens: collects the total light intensity after encoding the object to be measured; The third lens: turns the light beam into quasi-collimated light; The first flat beam splitter: splits the light into two beams, including reflected light and transmitted light, where the energy ratio of reflected light to transmitted light is 3:7; Second flat beam splitter: splits the transmitted light after passing through the first flat beam splitter into reflected light and transmitted light again, where the ratio of reflected light energy to lens light energy is 1:1 Cos filter: filters the reflected light after the second flat beam splitter; Sin filter: filters the transmitted light passing through the second flat beam splitter; The first single pixel detector collects the total light intensity after being reflected by the first flat beam splitter; The second single-pixel detector collects the total light intensity after being filtered by the Cos filter; The third single-pixel detector collects the total light intensity after being filtered by the Sin filter.

2. A single-pixel hyperspectral imaging method based on dual filters, characterized in that: The following steps are involved: Step 1, data collection stage: after the light source is irradiated on the digital micromirror device and passes through the built system, the total light intensity of the object to be measured without passing through the filter and the total light intensity after passing through the Cos filter and the Sin filter are obtained respectively; Step 2, image reconstruction stage: Use the compressed sensing algorithm to reconstruct the image of the object, and reconstruct the image of the object to be tested without the filter according to the total light intensity obtained in step 1. all And the images O after passing through the Cos filter and Sin filter respectively cos and image O sin ; Step 3, material spectrum analysis stage: calculate the phasor point coordinates corresponding to each pixel point in the reconstructed image in step 2, and obtain the phasor spectrum distribution diagram of the image. The specific expression is as follows: Where (x0, y0) is the spatial coordinate of any pixel in the reconstructed image, G(x0, y0) represents the G coordinate value at the position (x0, y0), S(x0, y0) represents the S coordinate value at the position (x0, y0), and O(x0, y0) is cos Represents image O cos The pixel value at position (x0, y0) in O(x0, y0) all Represents image O all The pixel value at position (x0, y0) in O(x0, y0) sin Represents image O sin The pixel value at position (x0, y0) in the image; Step 4, hyperspectral color image reconstruction stage; according to the phasor spectrum distribution diagram obtained in step 3, the characteristic spectrum corresponding to each pixel in the reconstructed image is obtained, and the color corresponding to the pixel is reconstructed according to the characteristic spectrum. The above operation is performed on each pixel in the reconstructed image to obtain a color image of the hyperspectral object.

3. The method for single-pixel hyperspectral imaging based on dual filters according to claim 2, characterized in that: The specific process of step 1 is as follows: S11, irradiating a light source on a digital micromirror device, generating a coding pattern using the digital micromirror device, optically encoding the object to be measured, and projecting the coding pattern onto the object to be measured through a first lens; S12, collecting the total light intensity from the encoded object through the second lens and the third lens; S13, using a first flat beam splitter and a second flat beam splitter to split the encoded total light intensity into three beams; S14, using a Cos filter and a Sin filter to filter the reflected light and the transmitted light passing through the second flat beam splitter respectively; S15. Build a single-pixel computational imaging system, use the first single-pixel detector to collect the total intensity of light after reflection by the first flat beam splitter, use the second single-pixel detector to collect the total intensity of light after filtering by the Cos filter, and use the third single-pixel detector to collect the total intensity of light after filtering by the Sin filter.

4. The method of single-pixel hyperspectral imaging based on dual filters according to claim 2, characterized in that: In S13, the encoded total light intensity is divided into three beams using the first flat beam splitter and the second flat beam splitter. Specifically, the collected light is divided into two beams by the first flat beam splitter, the intensity value of one beam is collected by the first single-pixel detector, and the other beam is divided into two beams again by the second flat beam splitter, and respectively collected by the second single-pixel detector and the third single-pixel detector.

5. The method of single-pixel hyperspectral imaging based on dual filters according to claim 2, characterized in that: The single-pixel computational imaging system is constructed by one or more of a differential single-pixel imaging algorithm, a compressed sensing algorithm, an iterative or non-iterative algorithm, a linear or nonlinear algorithm, and a deep learning algorithm.

Images