Color and near-infrared single-pixel imaging system and color calibration method thereof for reconstructed image

Abstract

The invention discloses a color and near-infrared single-pixel imaging system and a color calibration method thereof for a reconstructed image, and belongs to the field of optical imaging. The system specifically comprises a light source, a camera, a DMD, a plane mirror, a light splitting module, a single-pixel detector, an analog-to-digital conversion module and a computer, a light source which is transmitted / reflected by an object is converged to the DMD through the camera, two plane mirrors are respectively arranged at a + / -12-degree overturning angle, a + 12-degree reflection beam is reflected by the dichroscope to form a red light wave band, a transmission beam is reflected by the dichroscope to form a blue light wave band, a-12-degree reflection beam is reflected by the dichroscope to form a green light wave band, and a transmission beam is reflected by the dichroscope to form a near-infrared wave band. Light intensity signals are detected by four single-pixel detectors at the same time, RGB and near-infrared reconstructed images are obtained through calculation of an analog-to-digital conversion module, and a final image is formed through color calibration. According to the method, color and near-infrared single-pixel imaging is realized while the color image quality is ensured, and the method has important practical application value.

Description

Technical Field

[0001] The invention belongs to the field of optical imaging single pixel imaging, and in particular relates to a color and near-infrared single pixel imaging system and a color calibration method for reconstructing an image thereof. Background Art

[0002] Single-pixel imaging technology has broad prospects in applications in non-visible light bands such as spectral imaging, near-infrared imaging, and terahertz imaging, especially in security monitoring, medical imaging, remote sensing detection, and food safety testing.

[0003] As an emerging imaging method, single-pixel imaging technology uses a spatial light modulator to modulate the incident light field, combines the light intensity information collected by the single-pixel detector, and finally restores the image of the target object through a reconstruction algorithm. Compared with traditional area array imaging, single-pixel imaging technology can achieve multi-band imaging at a lower cost, especially in non-visible light bands such as near-infrared and terahertz.

[0004] Color imaging and near-infrared imaging are two important areas of modern imaging technology and play a key role in many application scenarios. Color imaging is mainly used to capture information within the visible spectrum and is widely used in daily visual equipment, monitoring systems, remote sensing and other fields. Near-infrared imaging is sensitive to the near-infrared spectrum that is invisible to the human eye and is suitable for specific applications such as night vision, medical diagnosis and agricultural monitoring.

[0005] However, in traditional imaging systems, color imaging and near-infrared imaging often require independent hardware systems, which increases the complexity and cost of the equipment.

[0006] Color and near-infrared single-pixel imaging systems face some technical challenges: First, existing systems mostly use a dispersed optical path design, which is large in size, complex in structure, and lacks compactness, making it difficult to adapt to application scenarios that have high requirements for device portability. Second, the energy utilization rate of existing spectroscopic solutions is not high, especially in multi-band imaging, the system energy utilization rate is low, affecting imaging quality and system efficiency. In summary, current technology makes it difficult to simultaneously achieve color and near-infrared single-pixel imaging in a highly compact integrated system.

[0007] Text-person retrieval technology is widely used in person identification, for example, to find lost children. When a specific text description of a person is given, the text-person retrieval model can find the corresponding image based on the input text to achieve person identification. Summary of the invention

[0008] In view of the shortcomings of the existing technology, a color and near-infrared single-pixel imaging system and a color calibration method for the reconstructed image are proposed, which can achieve color and near-infrared single-pixel imaging while ensuring the quality of color images, and has important practical application value.

[0009] The color and near-infrared single-pixel imaging system consists of a light source, a camera lens, a spatial light modulator DMD, a plane reflector, a light splitting module, a single-pixel detector, an analog-to-digital conversion module and a computer.

[0010] The plane reflector includes a first plane reflector and a second plane reflector; the light splitting module includes a first light splitting module and a second light splitting module; the single pixel detector includes a first single pixel photodetector, a second single pixel photodetector, a third single pixel photodetector and a fourth single pixel photodetector; the analog-to-digital conversion module includes a first AD converter, a second AD converter, a third AD converter and a fourth AD converter;

[0011] The light beam emitted by the light source is aimed at the object to be tested; the light beam reflected by the object to be tested is converged to the DMD through the camera lens; the computer is connected to the DMD and the analog-to-digital conversion module to generate the Hadamard mask pattern and send it to the DMD;

[0012] As a spatial light modulator, DMD consists of an imaging module, a spatial light modulation module and a detection module. The imaging module images the object illuminated by the light source on the DMD surface to ensure the object-image relationship. The spatial light modulation module modulates and encodes the image of the object in the spatial domain, and separates it in the spectral domain based on the spatial domain spectroscopic scheme. The detection module is responsible for synchronously collecting the spatial and spectral information of the modulated R, G, B and near-infrared four-way light, and then transmitting it to the computer for full-color single-pixel imaging image reconstruction, aiming to achieve synchronous imaging of color and near-infrared.

[0013] Specifically:

[0014] The DMD uses a digital micromirror array for spatial light modulation, modulates the light beam by controlling the flipping of the Hadamard mask pattern, and reflects the incident light at an angle of ±12°, providing two light paths that are incident on the first plane reflector and the second plane reflector respectively;

[0015] The first plane reflector guides the light beam to the first beam collecting lens, and vertically converges the reflected light beam onto the optical axis of the first light splitting module. The first light splitting module includes a first dichroic mirror and a first filter. After passing through the first dichroic mirror, the light beam simultaneously reflects and transmits the light beam, wherein the reflected light beam is in the red light band and is sent to the second single-pixel photodetector, and the transmitted light beam passes through the blue light band after passing through the first filter and is sent to the first single-pixel photodetector;

[0016] The first single-pixel photodetector collects the light intensity signal of blue light, converts the light intensity signal into a digital signal through the first AD conversion channel, and then sends it to the computer; the second single-pixel photodetector collects the light intensity signal of red light, converts the light intensity signal into a digital signal through the second AD conversion channel, and then sends it to the computer.

[0017] The second plane reflector guides the light beam to the second beam collecting lens, and converges the reflected light beam onto the optical axis surface of the second spectroscopic module, and the second spectroscopic module includes a second dichroic mirror; after the light beam passes through the second dichroic mirror, it simultaneously reflects and transmits the light beam, the reflected light beam is sent to the third single-pixel photodetector in the green light band, and the transmitted near-infrared light beam is sent to the fourth pixel photodetector; the third single-pixel photodetector collects the light intensity signal of the green light, converts the light intensity signal into a digital signal through the third AD conversion channel, and sends it to the computer, the fourth single-pixel photodetector collects the light intensity signal of the near-infrared light, converts the light intensity signal into a digital signal through the fourth AD conversion channel, and then sends it to the computer.

[0018] Finally, the computer correlates all the digital signals of the analog-to-digital conversion module with the Hadamard mask image, restores the image of the object to be tested, and realizes synchronous imaging of color and near-infrared.

[0019] The synchronized image I is reconstructed by the following formula:

[0020] I=P T ×S=P -1 ×S=P -1 ×P×o=o

[0021] The number of mask patterns projected by the DMD is M, and P is the modulation matrix composed of the M mask patterns loaded on the spatial light modulator. i ,…,p M ] T ; S is the measurement matrix: S=[s1,s2,…,s i ,…,s M ] T For each measurement, the light intensity signal s received by the single-point detector is i Expressed as: s i =p i ×O;

[0022] p i is the mask pattern: p i =[p i1 ,p i2 ,…,p iN ], where each element corresponds to the intensity of each point of the mask pattern.

[0023] The pixel size of the reconstructed image is N, which corresponds to the matrix size of each mask pattern; O is a vector containing N elements O = [o1, o2, ..., o N ] T , each element reflects the corresponding pixel value of the target object at the corresponding spatial position.

[0024] The color calibration method for reconstructing an image by the color and near-infrared single-pixel imaging system comprises the following specific steps:

[0025] Step 1: The light source emits a light beam to illuminate the black card, which is converged on the DMD through the camera lens to obtain the grayscale value of the first restored RGB and the grayscale value of the first infrared image;

[0026] The calculation formula is:

[0027]

[0028]

[0029] Among them, s R1i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the black card and the i-th measurement is performed; s B1i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the black card and the i-th measurement is performed; s G1i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the black card during the i-th measurement; s IR1i It represents the measurement matrix detected by the detector corresponding to the infrared band when the light beam emitted by the light source illuminates the black card during the i-th measurement; h i It represents the digital signal obtained by the computer from the single-pixel photodetector when the light beam emitted by the light source illuminates the color card and converges on the DMD through the camera lens, and the DMD displays the i-th Hadamard pattern. R1 , L G1 , L B1 are the grayscale values ​​of R, G, and B restored first; L IR1 is the gray value of the first infrared image;

[0030] Step 2: The light source emits a light beam to illuminate the white card, which is focused on the DMD through the camera lens to obtain the grayscale value of the second restored RGB;

[0031] The calculation formula is:

[0032]

[0033]

[0034] Among them, L R2 ,,L G2L B2 are the grayscale values ​​of R, G, and B restored in the second step respectively; R2i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed; s B2i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed; s G2i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed;

[0035] Step 3: The light source emits a light beam to illuminate the gray card, which is focused on the DMD through the camera lens to obtain the grayscale value of the third restored RGB;

[0036] The calculation formula is:

[0037]

[0038]

[0039] L R3 L G3 L B3 are the grayscale values ​​of the third restored RGB respectively; R3i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement; s B3i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement is performed; s G3i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement;

[0040] Step 4: The light source emits a light beam to illuminate the object to be measured, which is focused on the DMD through the camera lens to obtain the grayscale value of the fourth restored RGB;

[0041] The calculation formula is:

[0042]

[0043]

[0044]

[0045] L R4 L G4 L B4 The grayscale values ​​of the fourth restored RGB respectively; R4i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the object to be measured for the i-th measurement; s B4iIt represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the object to be measured for the i-th measurement; s G4i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the object to be measured during the i-th measurement;

[0046] Step 5: Use the grayscale value of the first restored RGB to calculate the first mean value of R, G, and B:

[0047]

[0048] L R1i is the gray value of the i-th pixel in the gray value of the first restored R; L G1i is the gray value of the i-th pixel in the gray value of the first restored G; L B1i is the gray value of the i-th pixel in the gray value of the first restored B;

[0049] Step 6: Perform black level compensation on the grayscale values ​​of the white card, the grayscale card and the object to be measured, and calculate the fifth, sixth and seventh restored RGB grayscale values ​​in combination with the first mean value;

[0050] The specific calculation is as follows:

[0051]

[0052] L R5 , L G5 , L B5 is the gray value of the restored image of the object to be measured after black level compensation, and the fifth restored gray value of R, G, B is obtained; L R6 , L B6 , L G6 is the gray value of the restored image of the white color card after black level compensation, and the sixth restored gray values ​​of R, G, and B are obtained; L R7 , L B7 , L G7 The grayscale value of the gray color card restored image after black level compensation is obtained to obtain the seventh restored R, G, and B grayscale values.

[0053] Step 7: Use the fifth and sixth restored RGB grayscale values ​​to perform light source non-uniformity correction to obtain the eighth restored R, G, B grayscale values;

[0054] L R8 =L R5 / L R6 , L G8 =L G5 / L G6 , L B8 =L B5 / L B6 ;

[0055] L R8 , L G8 , L B8 is the grayscale value of the restored image of the object to be measured after light source correction, and the eighth restored R, G, B grayscale values ​​are obtained.

[0056] Step 8: Calculate the second mean values ​​of R, G, and B using the seventh restored grayscale values ​​of R, G, and B;

[0057]

[0058] L R7i is the gray value of the i-th pixel in the gray value of the first restored R; L G7i is the gray value of the i-th pixel in the gray value of the first restored G; L B7i is the gray value of the i-th pixel in the gray value of the first restored B;

[0059] Step 9, using the second mean values ​​of R, G, and B to calculate the first gain of white balance correction;

[0060] Based on the second mean of the G channel, the ratio of the second mean of the G channel to the second mean of the R and B channels is calculated as the first gain of the white balance correction of the R and B channels; the calculation formula of the first gain is:

[0061]

[0062] Step 10: multiply the first gain by the eighth restored grayscale value of R, G, B to obtain the corrected ninth restored grayscale value of R, G, B:

[0063] L R9 =k R ×L R8 , L G9 =k G ×L G8 , L B9 =k B ×L B8

[0064] k R , k G , k B The first gain of R, G, B white balance correction; L R9 , L G9 , L B9 is the grayscale value of the grayscale image to be measured after white balance correction, and the ninth restored grayscale values ​​of R, G, and B are obtained.

[0065] Step 11: Fuse the ninth restored grayscale values ​​of R, G, and B to obtain a color image, thus completing the image reconstruction.

[0066] Ninth, the grayscale values ​​of R, G, and B are restored, that is, the pixel values ​​of each channel of the color reconstructed image are corrected respectively; each channel is corrected respectively, that is, the reconstruction of the three channels of the image R, G, and B is completed, and the pixel values ​​of the three channels are fused to complete the reconstruction of the color image.

[0067] The advantages of the present invention are:

[0068] The present invention discloses a color and near-infrared single-pixel imaging system, which integrates color and near-infrared single-pixel imaging based on spatial spectroscopy, improves the energy utilization rate and imaging quality of the system, and can simultaneously obtain color and near-infrared single-pixel images.

[0069] 2 The present invention discloses a color calibration method for reconstructing images of a color and near-infrared single-pixel imaging system, which effectively improves the color accuracy of color images through black level compensation, light source non-uniformity correction, and white balance correction, and ensures the multispectral imaging capability of the system. The method is suitable for application scenarios such as agricultural detection, security monitoring, and medical imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a structural schematic diagram of a color and near-infrared single-pixel imaging system of the present invention;

[0071] Figure 2 A flow chart of a method for color calibration of a reconstructed image by a color and near-infrared single-pixel imaging system of the present invention;

[0072] Figure 3 This is a diagram showing the effect of an experimental process of visible light color reconstruction image processing according to an embodiment of the present invention.

[0073] Among them, 1-camera body, 2-camera lens, 3-DMD, 4-first plane reflector, 5-first beam collecting lens, 6-second plane reflector, 7-second beam collecting lens, 8-light source, 9-first dichroic mirror, 10-second dichroic mirror, 11-first filter, 12-first single-pixel photodetector, 13-first AD converter, 14-second single-pixel photodetector, 15-second AD converter, 16-third single-pixel photodetector, 17-third AD converter, 18-fourth single-pixel photodetector, 19-fourth AD converter, 20-computer, 21-object to be measured. DETAILED DESCRIPTION

[0074] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only partial embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work should fall within the scope of protection of the present invention.

[0075] The present invention discloses a color and near-infrared single-pixel imaging system and a color calibration method for reconstructed images. A symmetrical dual-light path splitting structure is designed according to the working characteristics of a digital micromirror array, and a symmetrical plane reflector and a beam collecting lens are set; at the same time, a spatial splitting method and four single-pixel photodetectors are used to simultaneously collect spectral information and spatial information. The present invention can realize simultaneous imaging of full-color visible light and near-infrared bands. In addition, the light beam modulated by the digital micromirror array is reflected to the beam collecting lens by a plane reflector, and combined with the designed spatial splitting structure, not only the energy utilization rate of the system can be improved, but also compact color and near-infrared integrated single-pixel imaging can be realized. The integrated compact structure reduces the volume of the optical system while improving the stability and anti-interference ability of the system.

[0076] The color and near-infrared single-pixel imaging system, such as Figure 1 As shown, it consists of a light source, a camera lens, a spatial light modulator DMD, a plane reflector, a spectroscopic module, a single pixel detector, an analog-to-digital conversion module and a computer.

[0077] The light source transmitted or reflected by the object is converged on the DMD through the camera lens for modulation. The ±12° flip angle in the DMD corresponds to the mask pattern of the positive and negative modes. A plane reflector is placed at the ±12° flip angle of the DMD to reflect the light beam. The +12° reflected light beam passes through a dichroic mirror and the reflected light beam is in the red light band. The transmitted light beam passes through the blue light band after passing through a blue filter. The -12° reflected light beam passes through a dichroic mirror and the reflected light beam is in the green light band. The transmitted light beam is in the near-infrared band. Then, four single-pixel detectors simultaneously detect the four-way light intensity signals, and the analog-to-digital conversion module calculates the four reconstructed images of R, G, B and near-infrared. The reconstructed images of the three channels of R, G, and B are then color calibrated to form the final image. The color and near-infrared single-pixel imaging system in the present invention can achieve single-pixel imaging of color and near-infrared while ensuring the quality of color images, and has important practical application value.

[0078] The plane reflector includes a first plane reflector and a second plane reflector; the light splitting module includes a first light splitting module and a second light splitting module; the single pixel detector includes a first single pixel photodetector, a second single pixel photodetector, a third single pixel photodetector and a fourth single pixel photodetector; the analog-to-digital conversion module includes a first AD converter, a second AD converter, a third AD converter and a fourth AD converter;

[0079] The light beam emitted by the light source is aimed at the object to be tested; the light beam reflected by the object to be tested is converged to the DMD through the camera lens; the computer is connected to the DMD and the analog-to-digital conversion module to generate the Hadamard mask pattern and send it to the DMD;

[0080] As a spatial light modulator, DMD consists of an imaging module, a spatial light modulation module and a detection module. The imaging module images the object illuminated by the light source on the DMD surface to ensure the object-image relationship. The spatial light modulation module modulates and encodes the image of the object in the spatial domain, and separates it in the spectral domain based on the spatial domain spectroscopic scheme. The detection module is responsible for synchronously collecting the spatial and spectral information of the modulated R, G, B and near-infrared four-way light, and then transmitting it to the computer for full-color single-pixel imaging image reconstruction, aiming to achieve synchronous imaging of color and near-infrared.

[0081] Specifically:

[0082] The DMD uses a digital micromirror array for spatial light modulation, modulates the light beam by controlling the flipping of the Hadamard mask pattern, and reflects the incident light at an angle of ±12°, providing two light paths that are incident on the first plane reflector and the second plane reflector respectively;

[0083] The first plane reflector guides the light beam to the first beam collecting lens, and vertically converges the reflected light beam onto the optical axis of the first light splitting module. The first light splitting module includes a first dichroic mirror and a first filter. After passing through the first dichroic mirror, the light beam simultaneously reflects and transmits the light beam, wherein the reflected light beam is in the red light band and is sent to the second single-pixel photodetector, and the transmitted light beam passes through the blue light band after passing through the first filter and is sent to the first single-pixel photodetector;

[0084] The first single-pixel photodetector collects the light intensity signal of blue light, converts the light intensity signal into a digital signal through the first AD conversion channel, and then sends it to the computer; the second single-pixel photodetector collects the light intensity signal of red light, converts the light intensity signal into a digital signal through the second AD conversion channel, and then sends it to the computer.

[0085] The second plane reflector guides the light beam to the second beam collecting lens, and converges the reflected light beam onto the optical axis surface of the second spectroscopic module, and the second spectroscopic module includes a second dichroic mirror; after the light beam passes through the second dichroic mirror, it simultaneously reflects and transmits the light beam, the reflected light beam is sent to the third single-pixel photodetector in the green light band, and the transmitted near-infrared light beam is sent to the fourth pixel photodetector; the third single-pixel photodetector collects the light intensity signal of the green light, converts the light intensity signal into a digital signal through the third AD conversion channel, and sends it to the computer, the fourth single-pixel photodetector collects the light intensity signal of the near-infrared light, converts the light intensity signal into a digital signal through the fourth AD conversion channel, and then sends it to the computer.

[0086] Finally, the computer correlates all the digital signals of the analog-to-digital conversion module with the Hadamard mask image, restores the image of the object to be tested, and realizes synchronous imaging of color and near-infrared.

[0087] The synchronized image I is reconstructed by the following formula:

[0088] I=p T ×S=P -1 ×S=P -1 ×P×O=O

[0089] The number of mask patterns projected by the DMD is M, and P is the modulation matrix composed of the M mask patterns loaded on the spatial light modulator. i ,…,p M ] T ; S is the measurement matrix: S=[s1,s2,…,s i ,…,s M ] T For each measurement, the light intensity signal s received by the single-point detector is i Expressed as: s i =p i ×O;

[0090] p i is the mask pattern: p i =[p i1 ,p i2 ,…,p iN ], where each element corresponds to the intensity of each point of the mask pattern.

[0091] The pixel size of the reconstructed image is N, which corresponds to the matrix size of each mask pattern; O is a vector containing N elements O = [o1, o2, ..., o N ] T , each element reflects the corresponding pixel value of the target object at the corresponding spatial position.

[0092] For example, if the pixels of the reconstructed image are 2*2, N is 4, and the size of the mask pattern matrix is ​​2*2; where M is the number of mask patterns placed, that is, the number of times the spatial light modulation is performed and the number of pixel values ​​detected by the detector. When using orthogonal mask patterns, the orthogonality ensures that the object information obtained by each sampling is independent of each other. Therefore, when the mask pattern loaded on the spatial light modulator is represented as an N-dimensional vector, full sampling can be achieved, and M=N at this time;

[0093] The color calibration method for reconstructing an image by the color and near-infrared single-pixel imaging system is as follows: Figure 2 As shown, the specific steps are as follows:

[0094] Step 1: The light source emits a light beam to illuminate the black card, which is focused on the DMD through the camera lens. After detection by four single-pixel photodetectors, the computer receives four digital signals to obtain the grayscale value of the first restored RGB and the grayscale value of the first infrared image;

[0095] The calculation formula is:

[0096]

[0097]

[0098] Among them, s R1i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the black card and the i-th measurement is performed; s B1i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the black card and the i-th measurement is performed; s G1i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the black card during the i-th measurement; s IR1i It represents the measurement matrix detected by the detector corresponding to the infrared band when the light beam emitted by the light source illuminates the black card during the i-th measurement; h i It represents the digital signal obtained by the computer from the single-pixel photodetector when the light beam emitted by the light source illuminates the color card and converges on the DMD through the camera lens, and the DMD displays the i-th Hadamard pattern. R1 , L G1 , L B1 are the grayscale values ​​of R, G, and B restored first; L IR1 is the gray value of the first infrared image;

[0099] Step 2: The light source emits a light beam to illuminate the white card, which is focused on the DMD through the camera lens. After detection by four single-pixel photodetectors, the computer receives four digital signals to obtain the grayscale value of the second restored RGB.

[0100] The calculation formula is:

[0101]

[0102] Among them, L R2 ,,L G2 L B2 are the grayscale values ​​of R, G, and B restored in the second step respectively; R2i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed; s B2i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed; s G2i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed;

[0103] Step 3: The light source emits a light beam to illuminate the gray card, which is focused on the DMD through the camera lens. After detection by four single-pixel photodetectors, the computer receives four digital signals to obtain the grayscale value of the third restored RGB.

[0104] The calculation formula is:

[0105]

[0106] L R3 L G3 L B3 are the grayscale values ​​of the third restored RGB respectively; R3i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement; s B3i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement is performed; s G3i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement;

[0107] Step 4: The light source emits a light beam to illuminate the object to be measured, which is focused on the DMD through the camera lens. After detection by four single-pixel photodetectors, the computer receives four digital signals to obtain the grayscale value of the fourth restored RGB;

[0108] The calculation formula is:

[0109]

[0110] L R4 L G4 L B4 The grayscale values ​​of the fourth restored RGB respectively; R4i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the object to be measured for the i-th measurement; s B4i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the object to be measured for the i-th measurement; s G4i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the object to be measured during the i-th measurement;

[0111] Step 5: Use the grayscale value of the first restored RGB to calculate the first mean value of R, G, and B:

[0112]

[0113] L R1i is the gray value of the i-th pixel in the gray value of the first restored R; L G1i is the gray value of the i-th pixel in the gray value of the first restored G; LB1i is the gray value of the i-th pixel in the gray value of the first restored B;

[0114] Step 6: Perform black level compensation on the grayscale values ​​of the white card, the grayscale card and the object to be measured, and calculate the fifth, sixth and seventh restored RGB grayscale values ​​in combination with the first mean value;

[0115] The fourth restored R, G, B image is subtracted from the first mean value of R, G, B to obtain the grayscale value of the restored image of the target object to be measured after black level compensation, thereby obtaining the fifth restored R, G, B grayscale value;

[0116] The second restored R, G, B image is subtracted from the first mean value of R, G, B to obtain the grayscale value of the white color card restored image after black level compensation, and the sixth restored R, G, B grayscale value is obtained;

[0117] The third restored R, G, B image is subtracted from the first mean value of R, G, B to obtain the grayscale value of the gray color card restored image after black level compensation, thereby obtaining the seventh restored R, G, B grayscale value.

[0118] The specific calculation is as follows:

[0119]

[0120] L R5 , L G5 , L B5 is the gray value of the restored image of the object to be measured after black level compensation, and the fifth restored gray value of R, G, B is obtained; L R6 , L B6 , L G6 is the gray value of the restored image of the white color card after black level compensation, and the sixth restored gray values ​​of R, G, and B are obtained; L R7 , L B7 , L G7 The grayscale value of the gray color card restored image after black level compensation is obtained to obtain the seventh restored R, G, and B grayscale values.

[0121] Step 7: Use the fifth and sixth restored RGB grayscale values ​​to perform light source non-uniformity correction to obtain the eighth restored R, G, B grayscale values;

[0122] The one-to-one ratio of the fifth restored R, G, B grayscale values ​​to the sixth restored R, G, B grayscale values ​​is the grayscale value of the restored image of the object to be measured after light source correction, that is, the eighth restored R, G, B grayscale value.

[0123] L R8 =L R5 / L R6 , L G8 =L G5 / L G6 , L B8 =L B5 / L B6 ;

[0124] L R8 , L G8 , L B8 is the grayscale value of the restored image of the object to be measured after light source correction, and the eighth restored R, G, B grayscale values ​​are obtained.

[0125] Step 8: Calculate the second mean values ​​of R, G, and B using the seventh restored grayscale values ​​of R, G, and B;

[0126]

[0127] L R7i is the gray value of the i-th pixel in the gray value of the first restored R; L G7i is the gray value of the i-th pixel in the gray value of the first restored G; L B7i is the gray value of the i-th pixel in the gray value of the first restored B;

[0128] Step 9, using the second mean values ​​of R, G, and B to calculate the first gain of white balance correction;

[0129] Based on the second mean of the G channel, the ratio of the second mean of the G channel to the second mean of the R and B channels is calculated as the first gain of the white balance correction of the R and B channels; the first gain of R is obtained by dividing the second mean of R and G; the first gain of B is obtained by dividing the second mean of B and G; the first gain of G is 1; the calculation formula of the first gain is:

[0130]

[0131] Step 10: multiply the first gain by the eighth restored grayscale value of R, G, B to obtain the grayscale value of the grayscale image to be measured after white balance correction, that is, the ninth restored grayscale value of R, G, B:

[0132] L R9 =k R ×L R8 , L G9 =k G ×L G8 , L B9 =k B ×L B8

[0133] k R , k G , k B The first gain of R, G, B white balance correction; L R9 , L G9 , L B9is the grayscale value of the grayscale image to be measured after white balance correction, and the ninth restored grayscale values ​​of R, G, and B are obtained.

[0134] Step 11: Fuse the ninth restored grayscale values ​​of R, G, and B to obtain a color image, thus completing the image reconstruction.

[0135] Ninth, restore the grayscale values ​​of R, G, and B, that is, the pixel values ​​of each channel of the color reconstructed image after correction; correct each channel of the color reconstructed image separately, that is, complete the reconstruction of the three channels of the image R, G, and B, and fuse the pixel values ​​of the three channels to complete the reconstruction of the color image. While ensuring the color accuracy of the image, the color reproduction of the color image is effectively improved.

[0136] Example:

[0137] The present invention proposes a color and near-infrared single-pixel integrated imaging system based on spatial spectrometry, and combines it with an effective color single-pixel imaging color calibration method, which can improve the color accuracy of color single-pixel images while ensuring the quality of color images. The realization of color and near-infrared single-pixel imaging has important practical application value.

[0138] The system uses DMD as a spatial light modulator and realizes synchronous imaging of color and near-infrared through a dual optical path design. It is mainly used for multi-band imaging and can simultaneously obtain visible color and near-infrared images. The system achieves efficient optical path separation and collection by combining a digital micromirror array (DMD), a dichroic mirror and a filter splitting solution.

[0139] The technical solution of the present invention is described in detail below in conjunction with specific embodiments.

[0140] Specifically include: light source, camera lens, digital micromirror array (DMD), plane reflector, spectroscopic module, single pixel detector, analog-to-digital conversion module and computer.

[0141] A camera body, a camera lens located inside the camera body, a digital micromirror array (DMD), a first plane reflector, a first beam collecting lens, a second plane emitting mirror and a second beam collecting lens, a first dichroic mirror, a second dichroic mirror, a filter, a first single-pixel photodetector, a second single-pixel photodetector, a third single-pixel photodetector, a fourth single-pixel photodetector connected to the camera body via a sleeve, and a light source, a multi-channel AD converter and a computer located outside the camera body.

[0142] The light source is used to emit a light beam and aim at the target imaging object; a highly stable white light LED light source is used. The light source illuminates the target object, and the reflected light includes visible light (red, green, blue light) and near infrared light.

[0143] The camera lens adopts a standard Nikon lens, which is used to converge the light beam reflected by the target object onto the working surface of the DMD, ensuring high-quality focusing of the light, thereby improving the overall imaging clarity of the system. The object light is directly projected onto the DMD through the lens, reducing intermediate interference factors.

[0144] The computer is electrically connected to the DMD and the multi-channel AD converter, and is used to generate N Hadamard mask patterns and send them to the DMD and control the DMD and the AD converter to perform synchronous triggering and receiving signals;

[0145] The DMD spatial light modulator is the core component of the system. It is composed of tens of thousands of micromirrors that can modulate the incident light by ±12° flipping according to the computer control. The light reflected by the target object first passes through the camera lens and is focused onto the DMD surface; the DMD performs spatial modulation on the light according to the preset modulation pattern sent by the computer; through the ±12° flipping of the DMD, the light is divided into two optical paths: the +12° optical path and the -12° optical path.

[0146] A digital micromirror array (DMD) is used for spatial light modulation, which is used to modulate the light beam by controlling the flipping of the micromirrors with the N Hadamard mask patterns transmitted by the computer, and reflect the incident light at an angle of ±12°, providing two light paths to be incident on the first plane reflector and the second plane reflector respectively;

[0147] Plane reflector: used to guide the modulated light into the spectrometer module, ensuring that the light path is accurately transmitted to the spectrometer while maintaining the compact design of the system's light path.

[0148] The spectroscopic module is responsible for dividing the light modulated by the DMD into different spectral channels. The spectroscopic scheme of the present invention adopts a joint design of a dichroic mirror and a filter: the +12° optical path separates red light and blue light through the first dichroic mirror. The wavelength of red light is 625nm~700nm, and the wavelength of blue light is 410nm~490nm; the -12° optical path separates green light and near-infrared light through the second dichroic mirror. The wavelength of green light is 495nm~615nm, and the wavelength of near-infrared light is 670nm~1200nm. Compared with the traditional X-Cube solution, this spectroscopic design has a wider bandwidth in the visible light solution, higher energy utilization, and can obtain more spectral information.

[0149] The first plane reflector is used to guide the light beam to the first beam collecting lens; the first beam collecting lens is used to converge the light beam onto the first spectroscopic module. The first spectroscopic module divides the light beam into a blue light component with a transmission band of 410 to 490 nm and a red light component with a reflection band of 625 to 100 nm. The first single-pixel photodetector receives the converged blue light, collects its light intensity signal, converts the light intensity signal into a digital signal through the first AD conversion channel, and then sends it to the computer; the second single-pixel photodetector receives the converged red light, collects its light intensity signal, converts the light intensity signal into a digital signal through the second AD conversion channel, and then sends it to the computer. The angular bisector of the first plane reflector is the angular bisector of the light reflected by the micromirror on the DMD deflected by +12° and the light reflected by the first plane reflector to the first beam collecting lens.

[0150] The second plane reflector is used to guide the light beam to the second beam collecting lens; the second beam collecting lens is used to converge the light beam onto the second spectroscopic module. The second spectroscopic module divides the light beam into a near-infrared component with a transmission band of 670 to 1200 nm and a green component with a reflection band of 495 to 615 nm. The third single-pixel photodetector receives the converged green light, collects its light intensity signal, converts the light intensity signal into a digital signal through the third AD conversion channel, and then sends it to the computer. The fourth single-pixel photodetector receives the converged near-infrared light, collects its light intensity signal, converts the light intensity signal into a digital signal through the fourth AD conversion channel, and then sends it to the computer. The angular bisector of the second plane reflector is the angular bisector of the light reflected by the micromirror on the DMD deflected by -12° and the light reflected by the second plane reflector to the second beam collecting lens.

[0151] The first plane reflector and the second plane reflector both reflect the light beams, and the two reflected light beams are perpendicularly incident on the optical axis of the light splitting module, and the main rays of the reflected light beams coincide with the central axis of the light splitting module.

[0152] The first plane reflector and the second plane reflector are arranged symmetrically with respect to a symmetry plane, and the first beam collecting lens and the second beam collecting lens are also placed symmetrically with respect to the symmetry plane. The symmetry plane passes through the center point of the DMD, is parallel to the rotation axis of the DMD, and forms an angle of 45° with the bottom surface of the camera body.

[0153] The first spectroscopic module comprises a first dichroic mirror and a first filter. The first dichroic mirror transmits a wavelength band of 400 to 580 nm and reflects a wavelength band of 625 to 795 nm. A filter 1 is placed in the wavelength band transmitted by the first dichroic mirror.

[0154] The second light splitting module comprises a second dichroic mirror, the second dichroic mirror transmits 670 to 1200 nm and reflects 495 to 615 nm;

[0155] The single-pixel detector has an operating band of 340nm to 1100nm. The single-pixel detector collects light intensity signals one by one to provide raw data for subsequent image reconstruction.

[0156] The analog-to-digital conversion module is responsible for converting the analog light intensity signal received from the single-pixel detector into a digital signal for computer processing. The system uses a high-speed analog-to-digital conversion module with a sampling rate of 500kSPS to ensure efficient data acquisition, especially when the DMD is modulated at a higher frequency.

[0157] The computer performs correlation calculations on the digital signals converted by the first AD converter, the digital signals converted by the second AD converter, the digital signals converted by the third AD converter, and the digital signals converted by the fourth AD converter, respectively, and performs image restoration on the object to be measured.

[0158] The computer is responsible for the control of the entire system and image reconstruction. Specifically, the computer generates a modulation pattern based on the preset Hadamard matrix and projects it onto the target object through the DMD. At the same time, the computer controls the analog-to-digital conversion module to collect the light intensity signal of the detector. After accumulating multiple light intensity data, the computer uses a reconstruction algorithm to convert this data into the final color and near-infrared image. The resolution of the image reaches 128×128 pixels.

[0159] In order to ensure the color accuracy of color single-pixel imaging, the present invention proposes an improvement scheme based on black level compensation, light source correction and white balance correction for the color single-pixel imaging system based on spatial spectrometry. Due to the non-uniformity of the light source and the difference in the detector's response to light of different wavelengths, color distortion may occur during color imaging. To this end, the system performs the following processing on the image:

[0160] The system first images a black color plate with a single pixel, obtains the average value of the imaged pixels as the threshold for black level compensation, and then images a white paper with a single pixel to obtain the spatial intensity distribution information of the light source. According to the Retinex theory, white paper should present consistent light intensity under uniform illumination, so any spatial intensity difference can be regarded as the unevenness of the light source. The system records this light intensity distribution, and in subsequent image reconstruction, corrects the light intensity of each pixel to make the brightness distribution of the reconstructed image more uniform. White balance correction is used to eliminate the influence of different light source color temperatures on the imaging results. The system images a gray card, measures its average light intensity values ​​in the red, green, and blue channels, and calculates the white balance correction gain based on these measured values. The corresponding correction gain is applied to each channel of the color reconstructed image to ensure the color accuracy of the image. This method effectively improves the color reproduction of the color image.

[0161] like Figure 3 As shown, (a) is the unprocessed reconstructed image, (b) is the reconstructed image after white balance correction, and (c) is the reconstructed image after light source non-uniformity correction and white balance correction. The system of the present invention successfully realizes the simultaneous imaging of visible full color and near infrared. After black level compensation, light source non-uniformity correction and white balance correction, the color reproduction of the color image is significantly improved. The red, green and blue channels in the image are close to the actual object color, and the near infrared image can accurately capture the infrared spectrum information of the object.

[0162] In addition, through the differential imaging mode, the system can effectively remove additive noise and significantly improve the signal-to-noise ratio of the image. The final reconstructed image resolution is 128×128 pixels, and the imaging time is short, which is suitable for real-time multispectral imaging applications.

[0163] The present invention realizes simultaneous imaging of color and near-infrared through a compact optical path design and an efficient spectroscopic scheme, and performs color accuracy calibration on the reconstructed image to obtain color single-pixel imaging with high color reproduction.

Claims

1. A color and near-infrared single-pixel imaging system, characterized in that: Specifically include: It consists of a light source, a camera lens, a spatial light modulator DMD, a plane reflector, a spectroscopic module, a single pixel detector, an analog-to-digital conversion module and a computer; The plane reflector includes a first plane reflector and a second plane reflector; the light splitting module includes a first light splitting module and a second light splitting module; the single pixel detector includes a first single pixel photodetector, a second single pixel photodetector, a third single pixel photodetector and a fourth single pixel photodetector; the analog-to-digital conversion module includes a first AD converter, a second AD converter, a third AD converter and a fourth AD converter; The light beam emitted by the light source is aimed at the object to be measured; the transmitted or reflected light beam is converged to the DMD through the camera lens for modulation; at the same time, the computer is connected to the DMD and the analog-to-digital conversion module to generate the Hadamard mask pattern and send it to the DMD; The ±12° flip angle in the DMD corresponds to the mask pattern of the positive and negative modes. A plane mirror is placed at the ±12° flip angle of the DMD to reflect the light beam. The +12° reflected light beam passes through a dichroic mirror and the reflected light beam is in the red light band, and the transmitted light beam passes through the blue light band after passing through a blue filter. The -12° reflected light beam passes through a dichroic mirror and the reflected light beam is in the green light band, and the transmitted light beam is in the near-infrared band. Four single-pixel detectors detect the four light intensity signals simultaneously, and the analog-to-digital conversion module calculates the four reconstructed images of R, G, B and near-infrared. The reconstructed images of the three channels of R, G, and B are then color calibrated to form the final image.

2. A color and near-infrared single-pixel imaging system as claimed in claim 1, characterized in that: The DMD is a spatial light modulator, which is composed of an imaging module, a spatial light modulation module and a detection module; the imaging module images the object illuminated by the light source on the DMD surface to ensure the object-image relationship; The spatial light modulation module modulates and encodes the image of the object in the spatial domain, and separates it in the spectral domain based on the spatial domain spectroscopic scheme; the detection module is responsible for synchronously collecting the spatial and spectral information of the modulated R, G, B and near-infrared four-way light, and then transmitting it to the computer for full-color single-pixel imaging image reconstruction, aiming to achieve synchronous imaging of color and near-infrared.

3. A color and near-infrared single-pixel imaging system as claimed in claim 1, characterized in that: The DMD uses a digital micromirror array to perform spatial light modulation, modulates the light beam by controlling the flipping of the Hadamard mask pattern, reflects the incident light at an angle of ±12°, and provides two light paths that are incident on the first plane reflector and the second plane reflector respectively; The first plane reflector guides the light beam to the first beam collecting lens, and vertically converges the reflected light beam onto the optical axis of the first light splitting module. The first light splitting module includes a first dichroic mirror and a first filter. After passing through the first dichroic mirror, the light beam simultaneously reflects and transmits the light beam, wherein the reflected light beam is in the red light band and is sent to the second single-pixel photodetector, and the transmitted light beam passes through the blue light band after passing through the first filter and is sent to the first single-pixel photodetector; The first single-pixel photodetector collects the light intensity signal of blue light, converts the light intensity signal into a digital signal through the first AD conversion channel, and then sends it to the computer; the second single-pixel photodetector collects the light intensity signal of red light, converts the light intensity signal into a digital signal through the second AD conversion channel, and then sends it to the computer; The second plane reflector guides the light beam to the second light beam collecting lens, and converges the reflected light beam onto the optical axis surface of the second light splitting module, and the second light splitting module includes a second dichroic mirror; after the light beam passes through the second dichroic mirror, it simultaneously reflects and transmits the light beam, the reflected light beam is sent to the third single-pixel photodetector in the green light band, and the transmitted near-infrared light beam is sent to the fourth pixel photodetector; the third single-pixel photodetector collects the light intensity signal of the green light, converts the light intensity signal into a digital signal through the third AD conversion channel, and sends it to the computer, and the fourth single-pixel photodetector collects the light intensity signal of the near-infrared light, converts the light intensity signal into a digital signal through the fourth AD conversion channel, and then sends it to the computer; Finally, the computer correlates all the digital signals of the analog-to-digital conversion module with the Hadamard mask image, restores the image of the object to be tested, and realizes simultaneous imaging of color and near-infrared.

4. A color and near-infrared single-pixel imaging system as claimed in claim 3, characterized in that: The synchronized image I is reconstructed by the following formula: I=P T ×S=P -1 ×S=P -1 ×P×O=O The number of mask patterns projected by the DMD is M, and P is the modulation matrix composed of the M mask patterns loaded on the spatial light modulator. i ,…,p M ] T ; S is the measurement matrix: S=[s1,s2,…,s i ,…,s M ] T For each measurement, the light intensity signal s received by the single-point detector is i Expressed as: s i =p i ×P; p i is the mask pattern: p i =[p i1 ,p i2 ,…,p iN ], where each element corresponds to the intensity of each point of the mask pattern; The pixel size of the reconstructed image is N, which corresponds to the matrix size of each mask pattern; O is a vector containing N elements O = [o1, o2, ..., o N ] T , each element reflects the corresponding pixel value of the target object at the corresponding spatial position.

5. A method for color calibration of a reconstructed image using a color and near-infrared single-pixel imaging system as claimed in claim 1, characterized in that: The specific steps are as follows: Step 1: The light source emits a light beam to illuminate the black card, which is converged on the DMD through the camera lens to obtain the grayscale value of the first restored RGB and the grayscale value of the first infrared image; The calculation formula is: Among them, s R1i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the black card and the i-th measurement is performed; s B1i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the black card and the i-th measurement is performed; s G1i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the black card during the i-th measurement; s IR1i It represents the measurement matrix detected by the detector corresponding to the infrared band when the light beam emitted by the light source illuminates the black card during the i-th measurement; h i It represents the digital signal obtained by the computer from the single-pixel photodetector when the light beam emitted by the light source illuminates the color card and converges on the DMD through the camera lens, and the DMD displays the i-th Hadamard pattern. R1 , L G1 , L B1 are the grayscale values ​​of R, G, and B restored first; L IR1 is the gray value of the first infrared image; Step 2: The light source emits a light beam to illuminate the white card, which is focused on the DMD through the camera lens to obtain the grayscale value of the second restored RGB; The calculation formula is: Among them, L R2 ,,L G2 L B2 are the grayscale values ​​of R, G, and B restored in the second step respectively; R2i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed; s B2i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the white card and the i-th measurement is performed; s G2i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the white card and the i-th measurement; Step 3: The light source emits a light beam to illuminate the gray card, which is focused on the DMD through the camera lens to obtain the grayscale value of the third restored RGB; The calculation formula is: L R3 L G3 L B3 are the grayscale values ​​of the third restored RGB respectively; R3i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement; s B3i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement is performed; s G3i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the gray card and the i-th measurement; Step 4: The light source emits a light beam to illuminate the object to be measured, which is focused on the DMD through the camera lens to obtain the grayscale value of the fourth restored RGB; The calculation formula is: L R4 L G4 L B4 The grayscale values ​​of the fourth restored RGB respectively; R4i It represents the measurement matrix detected by the detector corresponding to the red light band when the light beam emitted by the light source illuminates the object to be measured for the i-th measurement; s B4i It represents the measurement matrix detected by the detector corresponding to the blue light band when the light beam emitted by the light source illuminates the object to be measured for the i-th measurement; s G4i It represents the measurement matrix detected by the detector corresponding to the green light band when the light beam emitted by the light source illuminates the object to be measured during the i-th measurement; Step 5: Use the grayscale value of the first restored RGB to calculate the first mean value of R, G, and B: L R1i is the gray value of the i-th pixel in the gray value of the first restored R; L G1i is the gray value of the i-th pixel in the gray value of the first restored G; L B1i is the gray value of the i-th pixel in the gray value of the first restored B; Step 6: Perform black level compensation on the grayscale values ​​of the white card, the grayscale card and the object to be measured, and calculate the fifth, sixth and seventh restored RGB grayscale values ​​in combination with the first mean value; The specific calculation is as follows: L R5 , L G5 , L B5 is the gray value of the restored image of the object to be measured after black level compensation, and the fifth restored gray value of R, G, B is obtained; L R6 , L B6 , L G6 is the gray value of the restored image of the white color card after black level compensation, and the sixth restored gray values ​​of R, G, and B are obtained; L R7 , L B7 , L G7 The grayscale value of the gray color card restored image after black level compensation is obtained to obtain the seventh restored grayscale value of R, G, and B; Step 7: Use the fifth and sixth restored RGB grayscale values ​​to perform light source non-uniformity correction to obtain the eighth restored R, G, B grayscale values; L R8 =L R5 / L R6 ,L G8 =L G5 / L G6 ,L B8 =L B5 / L B6 ; L R8 , L G8 , L B8 is the grayscale value of the restored image of the object to be measured after light source correction, and obtains the eighth restored grayscale value of R, G, B; Step 8: Calculate the second mean values ​​of R, G, and B using the seventh restored grayscale values ​​of R, G, and B; L R7i is the gray value of the i-th pixel in the gray value of the first restored R; L G7i is the gray value of the i-th pixel in the gray value of the first restored G; L B7i is the gray value of the i-th pixel in the gray value of the first restored B; Step 9, using the second mean values ​​of R, G, and B to calculate the first gain of white balance correction; Based on the second mean of the G channel, the ratio of the second mean of the G channel to the second mean of the R and B channels is calculated as the first gain of the white balance correction of the R and B channels; the calculation formula of the first gain is: k G ＝1， Step 10: multiply the first gain by the eighth restored grayscale value of R, G, B to obtain the corrected ninth restored grayscale value of R, G, B: L R9 =k R ×L R8 ,L G9 =k G ×L G8 ,L B9 =k B ×L B8 k R , k G , k B The first gain of R, G, B white balance correction; L R9 , L G9 , L B9 is the grayscale value of the grayscale image to be measured after white balance correction, and the ninth restored grayscale values ​​of R, G, and B are obtained; Step 11: Fusing the grayscale values ​​of R, G, and B restored in the ninth step to obtain a color image, thus completing the image reconstruction; Ninth, the grayscale values ​​of R, G, and B are restored, that is, the pixel values ​​of each channel of the color reconstructed image are corrected respectively; each channel is corrected respectively, that is, the reconstruction of the three channels of the image R, G, and B is completed, and the pixel values ​​of the three channels are fused to complete the reconstruction of the color image.

Images