Method, device and storage medium for measuring three-dimensional spectral intensity distribution of a target space

Abstract

The present invention discloses a method, apparatus, and storage medium for measuring the three-dimensional spectral intensity distribution of a target space. The measurement method includes: calibrating the measurement device; calibrating the radiation intensity of the imaging device; curve-fitting the function between the refocusing coefficient and the corresponding spatial depth; photographing a uniform white light source and recording the spectral light field white light source image; calculating the similarity between each sub-image in the white light source image and the central reference sub-image of the white light source image and the output correction coefficient of each sub-image; photographing and recording the spectral light field image of the target space, and correcting the spectral light field image of the target space; and calculating and forming the three-dimensional spectral intensity distribution field of the target space according to the corrected spectral light field image. The present invention obtains the spectral energy distribution at different focal depths in the target space, and then through all-focus calculation, obtains the spatial depth information and three-dimensional spectral energy distribution information of the object group in the target space, realizing the measurement of the three-dimensional multi-spectral intensity of the target space.

Description

Technical Field

[0001] The present invention relates to a device and method for measuring three-dimensional multi-spectral intensity distribution based on a single-spectral light field camera, belonging to the field of multi-spectral measurement Background Art

[0002] Multi-spectral imaging technology is a new type of technology formed by combining spectral technology and optical imaging technology. Currently, the mature multi-spectral imaging systems mainly include three categories: interferometric spectral imagers, tunable filter spectral imagers, and dispersive (prism / grating) spectral imagers

[0003] The characteristic of an interferometric spectral imager is to reconstruct the spectrum by using the interference fringes formed in each band and the Fourier transform relationship of the spectrum. According to the scanning principle, interferometric spectral imagers can be roughly divided into three types: time modulation type, spatial modulation type, and spatio-temporal hybrid modulation type. This type of imaging spectrometer requires a high-precision mechanical moving device to form interference to obtain spectral information, and is only applicable to the spectral image measurement of stationary two-dimensional in-focus targets, so its application field is greatly limited

[0004] The principle of a tunable filter spectral imager is to install a tunable filter with a tunable light passing band in front of the camera lens, so that the light radiation in the required band reaches the camera sensor. By adjusting the passable wavelength of the filter and performing multiple exposures at different bands, the multi-channel spectral image of the target object can be obtained. The principle of this spectral imaging system is simple and has good reliability. However, this system requires multiple exposures to complete the acquisition of image information. Therefore, the more bands are collected, the lower the time resolution of the system. On the other hand, when the field of view angle increases, the central wavelength and transmittance of the filter also change greatly. Therefore, it is greatly limited in the application of large field of view angles

[0005] The dispersive (prism / grating) spectral imager is the multi-spectral imaging system with the longest development history. According to different scanning methods, it mainly has two types: swing-scanning type and push-broom type. Its principle is that the light incident from the entrance slit is collimated by the collimation system, then dispersed by the grating or prism, and finally the imaging system images the slit at different positions of the pixels on the detector according to the wavelength, and completes the imaging of the entire target surface through scanning. It takes several seconds for the dispersive spectral imager to scan a target surface, which is not suitable for high-dynamic scenes. In addition, the one-dimensional slit in this system greatly reduces the light energy utilization rate, the imaging quality is not high, and there is also an inherent spectral line bending phenomenon, making the image quality blurred

[0006] Generally speaking, the existing multi-spectral imaging systems are based on traditional imaging modes and rely on spatial or spectral scanning mechanisms to obtain multi-spectral data. Therefore, the time resolution is not high, and only the spectral information at the focal plane position can be obtained, making it difficult to meet the time resolution requirements for spectral information acquisition in dynamic and complex scenes Summary of the Invention

[0007] The technical problem to be solved by the present invention is to address the disadvantages of the existing multi-spectral imaging technology, such as low time resolution in scanning imaging and only being able to obtain spectral information at the focal plane position. Therefore, a set of imaging system based on a single-spectral light field camera for three-dimensional spectral imaging of the target space without a scanning mechanism and with instantaneous imaging, as well as a measurement method for measuring the three-dimensional spectral distribution of the target space using this system, are proposed.

[0008] A measurement device for three-dimensional multi-spectral intensity distribution based on a single-spectral light field camera, comprising:

[0009] A spectral main lens for imaging the radiation rays in the target space and dispersing the image into spectral images in each spectral band. This lens includes a sheet filter array and a lens group composed of multiple lenses; the sheet filter array contains W spectral filters with different wavelengths, assembled at the equivalent principal plane of the lens group, for dispersing the collected image into spectral images in each spectral band.

[0010] A light field camera system for photographing and recording the light field information of the radiation rays in the target space. This light field camera system includes a black and white CCD camera, a 1:1 relay system, and a microlens array; the microlens array is located on the imaging surface of the spectral main lens, for dispersing and imaging the spectral images collected by the spectral main lens at corresponding positions of different pixels on the sensor of the black and white CCD camera; the relay system is located between the black and white CCD camera and the microlens array, for projecting the light intensity distribution on the rear focal plane of the microlens array onto the sensor surface of the CCD camera; the optical distance L MLA between the microlens array and the camera sensor is equal to the focal length of the microlens array.

[0011] A computer for storing the light field pictures obtained by the black and white CCD camera and using the computational imaging light field refocusing algorithm to obtain the three-dimensional spectral intensity distribution of the target space.

[0012] For the described measurement device, it is characterized in that the spectral main lens and the light field camera system are connected and kept coaxial through a square cage plate, cage rods, and fixing screws.

[0013] For the described measurement device, it is characterized in that the 1:1 relay system is connected by docking two fixed-focus lenses with the same structure through a double male ring. The focal length of the fixed-focus lens in the 1:1 relay system is 50 mm, and the maximum aperture is F1.4. When the fixed-focus lenses form the 1:1 relay system, the aperture is adjusted to F1.4 and focused at infinity to achieve the 1:1 image transfer effect of the 1:1 relay system.

[0014] The described measuring device is characterized in that the sensor array of the light field camera system is divided into I×J square sub-images according to the arrangement of the microlens array, and each square sub-image contains K 2 pixels. In the sub-image, the sub-image in the i-th row and the j-th column is marked as M (i，j) , and the pixel in the p-th row and the q-th column of this sub-image is marked as

[0015] A method for measuring the three-dimensional spectral intensity distribution of a target space by using the described measuring device is characterized in that the steps are as follows:

[0016] Step 1: Calibrate the measuring device; use the measuring device to capture an image of a uniform white light source, adjust the positional relationship between the microlens array and the spectral main lens along the fixed cage rod so that the sub-images in the real-time image collected by the computer are tangent to each other; adjust the installation angle of the black-and-white camera CCD in the light field camera system so that the arrangement of the sub-images is the same as the pixel arrangement.

[0017] Step 2: Calibrate the radiation intensity of the imaging device, and curve-fit the relationship between the gray-scale mean value of the image and the corresponding radiation intensity;

[0018] Step 3: Calibrate the depth of the imaging system, and curve-fit the function Depth function between the refocusing coefficient and the corresponding spatial depth;

[0019] Step 4: Use the measuring device to capture a uniform white light source and record the spectral light field white light source image (hereinafter referred to as the white image). According to the brightness, structure, and contrast of the image, calculate the similarity between each sub-image in the white light source image and the central reference sub-image of the white image. According to the similarity calculation result, delimit the range of valid sub-images and calculate the output correction coefficient of each sub-image.

[0020] Step 5: Use the measuring device to capture and record the original spectral light field image of the target space. Use the image correction coefficients of each sub-image in Step 3 to correct the original spectral light field image of the target space. According to the filter combination method in the chip filter array, divide the original spectral light field image into a total of W spectral light field monochromatic images of different wavelengths.

[0021] Step 6: For the monochromatic images of the spectral light fields at each wavelength, use the computational imaging light field refocusing transfer formula to calculate the refocused monochromatic images at various depths in the target space, obtaining a group of refocused monochromatic images. A total of N refocused monochromatic images can be obtained from the monochromatic images of the spectral light fields at each wavelength. Using the all-focus transfer formula, based on the refocused monochromatic images, calculate the depth index and the all-focus monochromatic spectral image of the target space, forming a three-dimensional spectral intensity distribution field of W different wavelengths in the target space.

[0022] In step 3, the refocusing coefficient is denoted as n, which is a real number not less than -1 and not greater than 1. The function Depth function(n) between it and the corresponding spatial depth Depth is fitted using the following formula:

[0023] Depth = Depth function(n) = a 1 ·n 3 +a 2 ·n 2 +a 3 ·n + a 4

[0024] In the formula, a 1 、a 2 、a 3 、a 4 are fitting coefficients, all of which are constants, and a 1 > 0.

[0025] The similarity ssim(M (i，j) , M Bench ) between each sub-image and the central reference sub-image of the white image in step 4 is determined using the following formula:

[0026]

[0027]

[0028] In the formula, represents the brightness of sub-image M (i，j) , represents the contrast of sub-image M (i，j) , represents the gray value of pixel .

[0029]

[0030]

[0031]

[0032]

[0033] In the formula, represents the luminance of the reference sub-image M Bench , represents the contrast of the reference sub-image M Bench , represents the comparison structure number of the sub-image M (i，j) and the reference sub-image M Bench , C 1 , C 2 , C 3 are all constants not less than 1; l(M (i，j) , M Bench ), c(M (i，j) , M Bench ), s(M (i，j) , M Bench ) respectively represent the luminance similarity, contrast similarity and structure similarity of the sub-image M (i，j) and the reference sub-image M Bench .

[0034] ssim(M (i，j) , M Bench ) = [l(M (i，j) , M Bench )] α ·[c(M (i，j) , M Bench )] β ·[s(M (i，j) , M Bench )] γ

[0035] In the formula, α, β, and γ are weight coefficients, all of which are positive integers not less than 2.

[0036] In the fourth step, the output correction coefficient of the sub-image is determined by the following formula:

[0037]

[0038] In the fifth step, the refocused monochromatic image with wavelength λ is determined by the following formula:

[0039]

[0040] In the formula, represents the sub-image at the i-th row and j-th column of the spectral light field monochromatic image with wavelength λ, F is the distance from the exit pupil of the spectral main lens to the sensor of the black and white CCD camera, n is the refocusing coefficient not less than -1 and not greater than 1, and there are a total of N (n 1 , n 2 ,..., n N), where λ represents the wavelength of the image; i' and j' are the row and column numbers of the sub-image M (i，j) corresponding to the refocused image.

[0041] In step five, the all-focus monochromatic spectral image with wavelength λ and the depth distribution map DM are generated from N refocused images by the following formula:

[0042]

[0043] LF ref-p = [ref-p(LF FS (1, 1)), ref-p(LF FS (1, 2)),..., ref-p(LF FS (i″, j″)),..., ref-p(LF FS (I′, J′))]

[0044] DM = Depth function(LF ref-p )

[0045] In the formula, represents the Hamiltonian differential operator, WD is the set operator calculation window; the number of rows and columns of the all-focus image LF FS are i″ and j″ respectively, with a total of I × J pixels. ref-p(LF FS (i″, j″)) represents the refocusing coefficient corresponding to the pixel at the i″-th row and j″-th column of the all-focus image LF FS .

[0046] The present invention uses a single imaging system to collect the spectral image of the target space. The imaging system described in the present invention can record not only the light intensity information of the target during a single exposure, but also the light direction information and spectral information. According to the radiation intensity calibration and depth information calibration in steps two and three of the measurement method, the collected spectral image can be converted into the spectral and direction information of the light in the target space collected.

[0047] According to the basic principle of the imaging system, the light rays emitted or reflected by a spatial object point are re-converged by the lens of the imaging system and projected onto the camera sensor array to form a light spot. The size of the projected light spot and the degree of energy concentration depend on the relative position relationship between the object point and the focal plane of the object space of the imaging system. The light spot projected by the object point on the focal plane is the smallest, and the degree of energy concentration of the light spot is the highest. Therefore, the spectral energy distribution of the object point located on the focal plane of the imaging system can be accurately reflected on the imaging sensor. The single-camera imaging system described in the present invention utilizes the spectral information and direction information existing in the light rays collected by a single exposure, uses the light field refocusing transfer method to form a light field equation set, and through integral calculation, makes the recorded spectral light intensity be re-integrated at different spatial depths, obtains the spectral energy distribution at different focal depths in the target space, and then through full-focus calculation, obtains the spatial depth information and three-dimensional spectral energy distribution information of the object group to be photographed in the target space, and realizes the measurement of three-dimensional multi-spectral intensity in the target space.

[0048] The imaging device of the present invention is different from common spectral imaging devices. It adopts a spectral main lens with a coupled chip filter array and couples a microlens array between the sensors. The imaging process of the imaging system is as follows: The light beam collected by the imaging system is first collected and converged by the spectral main lens, and is dispersed into a series of spectral light beams with different wavelengths by the chip filter array in the spectral main lens. When the spectral light beams are transmitted by the imaging system to the microlens array, they are secondarily imaged by the microlens array into the square sub-images. The sub-images cover a certain range of pixels on the sensor surface, and these pixels record the intensity distribution and direction distribution of the spectral light beams after secondary imaging. The imaging process and the recording of the intensity distribution and direction distribution of the light beam are all completed in one exposure. Therefore, compared with common spectral imaging devices, the spectral main lens and microlens array of the imaging system enable the imaging device to not only record the intensity of the light rays projected onto the sensor, but also record the direction and spectral information of the light rays, thereby realizing the recording of a five-dimensional (three direction dimensions, one spectral dimension, one intensity dimension) spectral light field under a single exposure.

[0049] Beneficial effects: Compared with the existing spectral imaging technology, the present invention has the following advantages:

[0050] (1) The spectral measurement method of the present invention is based on digital image processing technology and realizes the analysis of five-dimensional spectral light field information. Based on the spectral light field radiation transfer model, the three-dimensional spatial distribution and three-dimensional multi-spectral intensity distribution of the target scene are synchronously analyzed from the five-dimensional composite spectral light field dataset, realizing the measurement ability of the three-dimensional spectral intensity distribution of the target scene, avoiding the repeated calculation of multi-dimensional information, and greatly improving the information processing and extraction efficiency. The measurement method of the present invention fully considers the aberration problems existing in the spectral image acquisition device. Therefore, during the data analysis process, image brightness, structure, and contrast correction are integrated, avoiding information waste and potential errors caused by aberration. Compared with traditional technologies, the measurement results of the three-dimensional spectral intensity distribution are more accurate.

[0051] (2) During the imaging process of traditional multi-spectral imaging devices, line-by-line or band-by-band scanning is required, so the imaging time is long and the time resolution (i.e., the acquisition frame rate) is low. Compared with the above traditional devices, the measurement device of the present invention does not require scanning and can collect the complete five-dimensional multi-spectral information of the target space in a single exposure. The time resolution (i.e., the acquisition frame rate) is greatly improved and can be applied to the measurement of the spectral intensity distribution in high-speed dynamic scenes.

[0052] (3) The present invention only contains a single imaging device, and the device does not contain scanning machinery. It has a simple structure, is convenient to carry, install, and can be applied to various complex environments and industrial sites. Description of the Drawings

[0053] Figure 1 Structural diagram of the measurement device;

[0054] Figure 2 Structural diagram of the spectral main lens;

[0055] Figure 3 Optical path diagram of the measurement device;

[0056] Figure 4 Calibration diagram of the measurement device and its partial enlarged view;

[0057] Figure 5 Example diagram of the spectral light field collected by the measurement device and its partial enlarged view;

[0058] Figure 6 Calibration result of the depth curve of the refocusing coefficient;

[0059] Figure 7 Evaluation result of the sub-image similarity;

[0060] Figure 8 Comparison of the spectral light field images before and after correction (wavelength is 540nm);

[0061] Figure 9 Reconstructed depth distribution map;

[0062] Figure 10 Reconstructed spectral intensity distribution diagram (wavelength is 540nm);

[0063] Wherein: 1 - spectral main lens, 2 - light field camera system, 3 - computer, 4 - microlens array, 5 - 1:1 relay imaging system, 6 - black and white CCD camera, 7 - double positive ring, 8 - camera sensor, 9 - cage plate, 10 - cage rod, 11 - target space plate, 12 - object point, 13 - light ray 1, 14 - light ray 2, 15 - lens group, 16 - chip filter array, 17 - equivalent optical path of the measurement system, 18 - imaging plane of the spectral main lens. Specific implementation mode

[0064] The present invention will be further described below in conjunction with the accompanying drawings and examples. The following examples are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent modifications made by those skilled in the art to the present invention fall within the scope defined by the appended claims of this application.

[0065] The present invention provides a method for measuring the three-dimensional spectral intensity distribution of a target space, and the steps are as follows:

[0066] Step 1, calibration of the measurement device. The purpose of calibration is to make the microlens array 4 exactly located on the imaging plane 18 of the spectral main lens 1, and the relay imaging of the microlens array 4 through the 1:1 relay system 5 is conjugate with the camera sensor 8, and its imaging structure is as Figure 3 shown.

[0067] The measurement device is as Figure 1 , including:

[0068] A spectral main lens 1, which is used to image the radiation rays in the target space and disperse the image into spectral images in each spectral band. The structure of this lens 1 is as Figure 2 shown, and it includes a chip filter array 16 and a lens group 15 composed of multiple lenses; the chip filter array 16 contains W spectral filters with different wavelengths. In this example, W = 3, and the three spectral channels correspond to 460, 540, and 610 nm respectively: the chip filter array 16 is assembled at the equivalent principal plane of the lens group 15 and is used to disperse the collected image into spectral images in each spectral band.

[0069] A light field camera system 2 is used to capture and record the light field information of the radiation light in the target space. The light field camera system includes a microlens array 4, a 1:1 relay system 5, and a black and white CCD camera 6; the microlens array 4 is located on the imaging surface 18 of the spectral main lens 1, and is used to disperse and image the spectral image collected by the spectral main lens 1 at the corresponding positions of different pixels of the sensor 8 inside the black and white CCD camera 6; the relay system 5 is located between the black and white CCD camera 6 and the microlens array 4, and is used to project the light intensity distribution on the rear focal plane of the microlens array 4 onto the surface of the CCD camera sensor 8; the optical distance L between the microlens array 4 and the sensor 8 MLA is equal to the focal length of the microlens array 4. In this example, L MLA = 400 μm.

[0070] A computer 3 is used to store the light field pictures obtained by the black and white CCD camera 6, and to obtain the three-dimensional spectral intensity distribution of the target space by using the computational imaging light field refocusing algorithm.

[0071] The spectral main lens and the light field camera system are connected and kept coaxial through a square cage plate 9, cage rods 10, and fixing screws.

[0072] The 1:1 relay system 5 is formed by docking and connecting two fixed-focus lenses with the same structure through a double male ring 7. In this example, the focal length of the fixed-focus lens in the 1:1 relay system 5 is 50 mm, and the maximum aperture is F1.4. When the fixed-focus lenses form the 1:1 relay system 5, the aperture is adjusted to F1.4 and focused at infinity to achieve the 1:1 image transfer effect of the 1:1 relay system.

[0073] The sensor array of the light field camera system 2 is divided into I×J square sub-images according to the arrangement mode of the microlens array 4, and each square sub-image contains K 2 pixels. In this case, a total of 186×135 square sub-images are included, and each sub-image contains 18 2 pixels. In the sub-image, the sub-image in the i-th row and j-th column is marked as M (i，j) , and the pixel in the p-th row and q-th column of this sub-image is marked as

[0074] The specific operation of calibrating the measuring device is as follows:

[0075] 1) First, use the computer 3 to set the camera 6 to the continuous acquisition - real-time display mode;

[0076] 2) Set the aperture of the spectral main lens 1 to F4 and set the focus ring to the nearest focus mode;

[0077] 3) Set up a uniform white light source in front of the measured main lens, and adjust the white light source so that its light-emitting surface is perpendicular to the measurement and is at the focal plane position of the measurement system;

[0078] 4) Use the measuring device to capture an image of the uniform white light source, and adjust the positional relationship between the microlens array 5 and the spectral main lens 1 along the fixed cage rod 10, so that the sub-images in the real-time image collected by the computer 3 are tangent to each other, as Figure 1 shown;

[0079] 5) Rotate the connection between the camera 7 and the cage plate 10, and adjust the installation angle of the black and white CCD camera 7 in the light field camera system so that the arrangement of the sub-images is the same as the pixel arrangement;

[0080] 6) Fix all the devices through the cage plate 9 and the cage plate 10, and the adjustment is completed.

[0081] Step 2: Calibrate the radiation intensity of the imaging device, and curve-fit the relationship between the average image gray value and the corresponding radiation intensity.

[0082] The specific steps are as follows:

[0083] 1) Place the calibrated measuring device in front of the blackbody furnace, fix the device and measure the distance between the spectral main lens 1 and the opening of the blackbody furnace;

[0084] 2) Set the temperature of the blackbody furnace to 500 degrees Celsius;

[0085] 3) Use the computer 3 to set the exposure time of the camera 7 so that the gray value rate of the image it captures is approximately equal to 15;

[0086] 4) Starting from the blackbody furnace temperature of 500 degrees Celsius, gradually increase the temperature of the blackbody furnace, and take an image of the blackbody furnace every 25 degrees Celsius until the temperature of the blackbody furnace reaches 1300 degrees Celsius;

[0087] 5) Fit a curve with the average gray value of the captured blackbody furnace images and the temperature points to obtain the conversion relationship between the average image gray value and the corresponding radiation intensity.

[0088] Step 3: Calibrate the depth of the imaging system, and curve-fit the function Depth function between the refocusing coefficient and the corresponding spatial depth.

[0089] The specific steps are as follows:

[0090] 1) Fix the calibrated measuring device on the optical platform, and set the focusing ring of the spectral main lens 1 to the nearest focus mode;

[0091] 2) Place a 1000 - mm long transparent ruler with scale in front of the main spectral lens 1, and adjust the angle of the ruler so that the scale surface of the ruler forms an angle of 30 degrees with the measuring device;

[0092] 3) Move the measuring device forward and backward so that the focal plane of the main spectral lens 1 is at the 500 - mm scale of the ruler;

[0093] 4) Use the measuring device of the present invention to take a light - field photo;

[0094] 5) According to the ruler angle, calculate the distance between each scale on the ruler and the 500 - mm scale, and convert it into spatial depth, that is, the distance between each scale and the main spectral lens;

[0095] 6) Perform refocusing calculation on the collected photo, fit the functional relationship between each refocusing coefficient and the spatial depth, and obtain the function Depth function between the refocusing coefficient and the corresponding spatial depth. In this example, the fitting formula is as follows:

[0096] Depth = Depth function(n)=a 1 ·n 3 +a 2 ·n 2 +a 3 ·n + a 4 (1)

[0097] In the formula, a 1 、a 2 、a 3 、a 4 are fitting coefficients, all of which are constants, and a 1 >0. In this example, the fitting results are a 1 =21.67, a 2 =13.61, a 3 =-15.24, a 4 =22.54, and the fitting curve is as Figure 6 shown

[0098] Step 4: Use the measuring device to photograph a uniform white light source and record the spectral light - field white - light - source image (hereinafter referred to as the white image). According to the brightness, structure, and contrast of the image, calculate the similarity between each sub - image in the white - light - source image and the central reference sub - image of the white image. According to the similarity calculation result, delimit the range of valid sub - images and calculate the output correction coefficient of each sub - image.

[0099] The similarity ssim(M (i，j) , M Bench ) between each sub - image and the central reference sub - image of the white image is determined by the following formula:

[0100]

[0101]

[0102] In the formula, represents the pixel gray value.

[0103]

[0104]

[0105]

[0106]

[0107] In the formula, represents the brightness of the reference sub-image M Bench , represents the contrast of the reference sub-image M Bench , represents the comparison structure number between the sub-image M (i，j) and the reference sub-image M Bench , C 1 , C 2 , C 3 are all constants not less than 1; l(M (i，j) , M Bench ), c(M (i，j) , M Bench ), s(M (i，j) , M Bench ) respectively represent the brightness similarity, contrast similarity and structure similarity between the sub-image M (i，j) and the reference sub-image M Bench .

[0108] ssim(M (i，j) , M Bench ) = [l(M (i，j) , M Bench )] α ·[c(M (i，j) , M Bench )]β·[s(M (i，j) , M Bench )] γ (8) In the formula, α, β, and γ are weight coefficients, all of which are positive integers not less than 2. In this example, α = 2.15, β = 2.26, and γ = 3. The distribution of the similarity calculation results between each sub-image and the central reference sub-image of the white image is as Figure 6 shown.

[0109] Output correction coefficient of the sub-image It is determined by the following formula:

[0110]

[0111] Step 5: Use the measuring device to capture and record the original spectral light field image of the target space. Use the image correction coefficients of each sub-image in Step 3 to correct the original spectral light field image of the target space. In this example, the comparison of the white image before and after correction at 540 nm is as Figure 8 shown. The brightness and contrast of the dark area at the edge of the corrected image are improved. In the following subsequent steps, the corrected original spectral light field image is used as the processing object.

[0112] According to the filter combination method in the chip filter array, the corrected original spectral light field image is segmented into a total of W = 3 corrected spectral light field monochromatic images of different wavelengths.

[0113] The refocused monochromatic image with wavelength λ is determined by the following formula:

[0114]

[0115] In the formula, represents the sub-image at the i-th row and j-th column of the corrected spectral light field monochromatic image with wavelength λ. F is the distance from the exit pupil of the spectral main lens to the sensor of the black and white CCD camera. n is the refocusing coefficient not less than -1 and not greater than 1, with a total of N (n 1 , n 2 ,..., n N ), and λ represents the wavelength of the image; i' and j' are the row number and column number of the sub-image M (i，j) corresponding to the refocused image. In this example, N = 25.

[0116] Step 6: For each corrected spectral light field monochromatic image of each wavelength, use the computational imaging light field refocusing transfer formula to calculate the refocused monochromatic images at various depths in the target space, obtaining a group of refocused monochromatic images. A total of N = 25 refocused monochromatic images can be obtained for each spectral light field monochromatic image of each wavelength. Using the full-focus transfer formula, based on the refocused monochromatic images, calculate the depth index and the full-focus monochromatic spectral image of the target space, forming a three-dimensional spectral intensity distribution field of W = 3 different wavelengths in the target space.

[0117] The full-focus monochromatic spectral image with wavelength λ and the depth distribution map DM are generated from N = 25 refocused images by the following formula:

[0118]

[0119] LFref-p = [ref - p(LF FS (1, 1)), ref - p(LF FS (1, 2)),..., ref - p(LF FS (i″, j″)),..., ref - p(LF FS (I′, J′))] (12)

[0121] DM = Depth function(LF ref - p)

[0122] In the formula, represents the Hamiltonian differential operator, WD is the set operator calculation window, in this example WD = 3×3 pixels; the number of rows and columns of the full - focus image LF FS are i″ and j″ respectively. ref - p(LF FS (i″, j″)) represents the refocusing coefficient corresponding to the pixel in the i″ - th row and j″ - th column of the full - focus image LF FS . In this example, the reconstructed three - dimensional space depth distribution and the full - focus spectral intensity distribution of the spectral light field are as shown in Figure 9 and 10 respectively.

Claims

1. A method for measuring the three-dimensional spectral intensity distribution of a target space, characterized in that, it includes: Step 1, calibration of the measuring device: making the sub-images in the real-time image collected tangent to each other; adjusting the installation angles of the CCDs of the black-and-white cameras in the light field camera system so that the arrangement of the sub-images is the same as the pixel arrangement; Step 2, calibration of the radiation intensity of the imaging device: curve-fitting the relationship between the average image gray value and the corresponding radiation intensity; the imaging device uses a spectral main lens with a coupled chip filter array and a microlens array is coupled between the sensors; Step 3, depth calibration of the imaging device; curve-fitting the function between the refocusing coefficient and the corresponding spatial depth; Step 4, using the calibrated measuring device, photographing a uniform white light source and recording the spectral light field white light source image; calculating the similarity between each sub-image in the white light source image and the central reference sub-image of the white light source image according to the brightness, structure and contrast of the white light source image; calculating the output correction coefficient of each sub-image according to the similarity calculation result; Step 5, using the calibrated measuring device, photographing and recording the spectral light field image of the target space; correcting the spectral light field image of the target space using the image correction coefficients of each sub-image in Step 4; Step 6, calculating to form a three-dimensional spectral intensity distribution field of the target space according to the corrected spectral light field image.

2. The method according to claim 1, characterized in that, Step 6 includes: Adopting a computational imaging light field refocusing algorithm to calculate the refocused images of the target space at various depths, obtaining a group of refocused images, a total of N; Adopting a full-focus algorithm to calculate the gray-scale gradients of each pixel on each refocused image and marking the gray scale and the corresponding refocusing coefficient of the pixel when the maximum gradient occurs; Combining the pixels with the maximum gradient to form a full-focus image of the full-focus spectral intensity distribution; Arranging the refocusing coefficients of each pixel at the maximum gradient according to the pixel arrangement combination to form a refocusing coefficient distribution map, and converting it into a depth distribution map according to the function marked in Step 3; Combining the full-focus image and the depth distribution map to form a three-dimensional spectral intensity distribution field of the target space.

3. The method according to claim 1, characterized in that: In Step 3, the refocusing coefficient is denoted as n, which is a real number not less than -1 and not greater than 1, and the function Depth function(n) between it and the corresponding spatial depth Depth is fitted by the following formula: Depth function(n)=a 1 ·n 3 +a 2 ·n 2 +a 3 ·n+a 4 Wherein, a 1 , a 2 , a 3 , a 4 are fitting coefficients, all being constants, and a 1 > 0.

4. The method according to claim 1, characterized in that: In Step 4, the similarity between each sub-image in the white light source image and the central reference sub-image of the white light source image is: ssim(M (i,j) ,M Bench ) = [l(M (i,j) ,M Bench )] α ·[c(M (i,j) ,M Bench )] β ·[s(M (i,j) ,M Bench )]γ In the formula, ssim(M (i,j) ,M Bench ) is the similarity between the sub-image M (i,j) at the i-th row and j-th column in the white light source image and the reference sub-image M Bench at the center of the white light source image. There are a total of I×J sub-images in the white light source image; α, β, and γ are weight coefficients, all of which are positive integers not less than 2; l(M (i,j) ,M Bench ) represents the luminance similarity between the sub-image M (i,j) and the reference sub-image M Bench ; c(M (i,j) ,M Bench ) represents the contrast similarity between the sub-image M (i,j) and the reference sub-image M Bench ; s(M (i,j) ,M Bench ) represents the structural similarity between the sub-image M (i ,j) and the reference sub-image M Bench .

5. The method according to claim 4, characterized in that: Luminance similarity l(M (i,j) ,M Bench ), contrast similarity c(M (i,j) ,M Bench ), and structure similarity s(M (i,j) ,M Bench are respectively as follows: In the formula, represents the brightness of the sub-image M (i,j) ; represents the contrast of the sub-image M (i,j) ; represents the brightness of the reference sub-image M Bench ; represents the contrast of the reference sub-image M Bench ; represents the comparison structure number between the sub-image M (i,j) and the reference sub-image M Bench , C 1 , C 2 , C 3 are all constants not less than 1.

6. The method according to claim 5, characterized in that: Sub-image M (i,j) brightness is as follows: Sub-image M (i,j) has a contrast of as follows: Sub-image M (i,j) The comparison structure number with the reference sub-image M Bench is as follows is: In the formula, represents the pixel at the p-th row and q-th column of the sub-image M (i,j) , and represents the pixel at the p-th row and q-th column of the reference sub-image M Bench ; represents the gray value of the pixel , and represents the gray value of the pixel ; each sub-image and the reference sub-image are square and cover K 2 pixels, where K represents the number of pixels covered by each side of the square.

7. The method according to claim 6, characterized in that: In Step 4, the calculated output correction coefficient of each sub-image is: In the formula, is the output correction coefficient of the sub-image.

8. The method according to claim 7, other features being ： In step six, the obtained refocused image is determined by the following formula: In the formula, is the refocused image; F is the distance from the exit pupil of the spectral main lens to the sensor of the black and white CCD camera; n is a refocusing coefficient not less than -1 and not greater than 1, with a total of N (n 1 , n 2 , …, n N ); i' and j' are the row number and column number of the sub-image M (i,j) corresponding to the refocused image; Full-focus image LF FS and refocusing coefficient index map LF ref-p and depth distribution map DM are generated from N refocused images by the following formula: DM = Depth function(LF ref-p ) LF ref-p = [ref-p(LF FS (1,1)), ref-p(LF FS (1,2)), …, ref-p(LF FS (i”, j”)), …, ref -p(LF FS (I',J'))] In the formula, represents the Hamiltonian differential operator, and WD is the set operator calculation window; the full-focus image LF FS has the number of rows and columns as i” and j” respectively, with a total of I×J pixels; ref-p(LF FS (i”,j”)) represents the refocusing coefficient corresponding to the pixel at the i”th row and j”th column on the full-focus image LF FS .

9. A device for measuring the three-dimensional spectral intensity distribution of a target space, characterized in that: it includes: An image acquisition device for photographing and recording the original spectral light field image of the target space; One or more processors; A memory for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method for measuring the three-dimensional spectral intensity distribution of the target space as described in any one of claims 1-8.

10. A storage medium having a computer program stored thereon, Characterized in that, When the program is executed by a processor, it implements the method for measuring the three-dimensional spectral intensity distribution of the target space as described in any one of claims 1-8.

Images