A micro hyperspectral camera and a measurement method

Abstract

The present invention discloses a micro hyperspectral camera and a measurement method, which relate to the fields of computational imaging and spectral measurement. The micro hyperspectral camera includes a band-pass filter for filtering out stray light outside the detection range and eliminating the influence of stray light on the measurement accuracy; an imaging lens for imaging the detection object at the diaphragm plane; a diaphragm for restricting the size of the field of view to ensure that the dispersed images do not overlap; a relay lens for focusing the image on the diaphragm onto the sensor; a dispersion element for dispersing the formed image in the spectral dimension to form a plurality of different images; and a sensor for collecting the dispersed image. The band-pass filter is fixed on the imaging lens and placed at the forefront. The imaging lens, the diaphragm, the relay lens, the dispersion element and the sensor are arranged in sequence front to back on the same axis. By reducing the number of optical elements and using a dispersion element with a large diffraction angle to reduce the longitudinal size of the spectral camera, the present invention reduces the volume and has both high spatial resolution and spectral resolution.

Description

Technical Field

[0001] The present invention relates to the fields of computational imaging and spectral measurement, and particularly to a miniature hyperspectral camera and a measurement method. Background Art

[0002] Spectral measurement has important applications in daily life and production. By detecting the spectral signals of the skin, real human faces can be distinguished from silicone head models, improving the accuracy of face recognition; by detecting the spectral signals in food, it can be determined whether the food has deteriorated; by detecting spectral signals through satellites, it can be determined whether there is a fire in the forest. Conventional spectral detection uses a spectrometer, which does not have spatial resolution and can only measure the spectral signals of a single point. At present, the main spectral measurement means with spatial resolution are as follows. One is to add narrow-band filters with different bands in front of a monochromatic camera, and the spectral information of an object is obtained through continuous measurement. However, this method does not have time resolution and can only be used for measuring time-invariant objects. The second is to perform measurement through a dispersive element and a coding element, called CASSI (Coded aperture snapshot spectral imaging). This method first disperses an object using a dispersive element, then spatially encodes the dispersed light, and the two-dimensional spectral information of the object can be restored by combining the principle of compressive sensing. The third is a reconstruction method that combines a dispersive element and the principle of tomography, called the CTIS (Computed Tomography Image Spectrometers) method. The conventional CTIS method requires collimated light to pass through a dispersive element and then be imaged through an imaging lens. Both the conventional CASSI and CTIS methods can achieve snapshot measurements, but these two methods require more optical elements and the device is complex. Therefore, a common defect of the above methods is that their devices are relatively large, which is not conducive to carrying and application in daily life. The miniaturization of spectral measurement is an important task in production practice.

[0003] Therefore, the technicians in this field are committed to developing a miniaturized spectral camera to make it convenient for people to carry around. Summary of the Invention

[0004] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is to miniaturize the spectral camera to make it convenient for people to carry around.

[0005] To achieve the above object, the present invention provides a miniature hyperspectral camera, including

[0006] a band-pass filter that filters out stray light outside the detection range and eliminates the influence of stray light on the measurement accuracy;

[0007] An imaging lens that images the detection object on the diaphragm plane;

[0008] A diaphragm that limits the size of the field of view so that the dispersed images do not overlap;

[0009] A relay lens that focuses the image on the diaphragm onto the sensor;

[0010] A dispersion element that disperses the formed image in the spectral dimension to form several different images;

[0011] A sensor that acquires the dispersed image;

[0012] The band-pass filter is fixed on the imaging lens, the band-pass filter is placed at the forefront, and the imaging lens, diaphragm, relay lens, dispersion element and sensor are arranged in sequence front and back on the same axis.

[0013] In a preferred embodiment of the present invention, the imaging lens and the relay lens are micro short-focus lenses, with focal lengths of 6 mm and 4 mm respectively, and their diameters are both less than 6 mm.

[0014] In a preferred embodiment of the present invention, the diaphragm is an adjustable micro diaphragm, its light-transmitting aperture is a circular hole, and the light-transmitting diameter is 0-2 mm.

[0015] In a preferred embodiment of the present invention, the dispersion element is a DOE with a large diffraction angle, a two-dimensional grating or a metasurface that is designed to divide a beam of light into several different beams of light. From the center to the edge, they are the 0th order, 1st order, 2nd order... The diffraction angles of different wavelengths of light are different after diffraction.

[0016] In a preferred embodiment of the present invention, the sensor is a micro monochrome CCD, its imaging area is 2.4 mm × 2.4 mm, and the output data is 10-bit binary numbers.

[0017] The present invention also provides a hyperspectral imaging method, which uses the micro hyperspectral camera and includes the following steps:

[0018] Step 1: Calibrate the hyperspectral camera using a monochromator, record the wavelength and intensity of the monochromator, and simultaneously record the image on the sensor;

[0019] Step 2: Calculate the response of the monochromatic light emitted from any point in space on the sensor according to the calibration data, and establish the correlation matrix therein as the imaging model;

[0020] Step 3: After completing the calibration and model establishment, image the actual object;

[0021] Step 4: Use the MLEM-TV algorithm combined with the guided image filtering algorithm to reconstruct the hyperspectral image, where the guidance image is the 0th order image after diffraction.

[0022] Preferably, the dynamic measurement of the hyperspectral image is achieved by repeatedly executing Step 3 to Step 4.

[0023] Preferably, Step 2 further includes:

[0024] Step 2.1: Assume that the size of the data cube of the object to be reconstructed after discretization is I×J×K, where I and J are the spatial dimensions and K is the spectral dimension;

[0025] Step 2.2: For the first wavelength, the nine images formed are respectively denoted as P11_1, P12_1, P13_1, P21_1, P22_1, P23_1, P31_1, P32_1, P33_1, where P22_1 is the 0th order image of the first wavelength, and the rest are the 1st order images of the first wavelength.

[0026] Step 2.3: After traversing all wavelengths, nine images are obtained.

[0027] Preferably, the spectral resolution of the monochromator is less than 1 nm.

[0028] Preferably, the reconstruction algorithm in Step 4 is an iterative method, and in each iteration, MLEM reconstruction, TV regularization, and guided image filtering are performed in sequence.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. The collimating lens in the CTIS system is removed, reducing the number of optical elements, decreasing the volume, and greatly reducing the production cost, and it can be mass-produced;

[0031] 2. A dispersive element (diffractive of element, DOE) with a large diffraction angle is designed, which can reduce the longitudinal size of the hyperspectral camera, greatly reduce the volume of the system, and is convenient to carry and use daily;

[0032] 3. The present invention uses a micro adjustable aperture to limit the field of view;

[0033] 4. The present invention has both high spatial resolution (1 mm) and spectral resolution (minimum 1 nm).

[0034] The following will further illustrate the concept, specific structure and technical effects generated by the present invention with reference to the accompanying drawings to fully understand the purpose, features and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1It is a schematic diagram of the optical path structure of a preferred embodiment of the present invention;

[0036] Figure 2 It is a photo of a circular light spot collected in a preferred embodiment of the present invention;

[0037] Figure 3 They are images of each wavelength after hyperspectral image reconstruction in a preferred embodiment of the present invention;

[0038] Figure 4 It is the spectrum measured in a preferred embodiment of the present invention;

[0039] Wherein, 1. Band-pass filter; 2. Imaging lens; 3. Diaphragm; 4. Relay lens; 5. Dispersion element; 6. Sensor. Detailed implementation manners

[0040] The following introduces multiple preferred embodiments of the present invention with reference to the accompanying drawings of the specification to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0041] In the drawings, components with the same structure are denoted by the same numerical labels, and components with similar structures or functions everywhere are denoted by similar numerical labels. The size and thickness of each component shown in the drawings are arbitrarily shown, and the present invention does not limit the size and thickness of each component. To make the illustration clearer, the thickness of some parts in the drawings is appropriately exaggerated.

[0042] As Figure 1 shown, a preferred embodiment of the present invention provides a micro-hyperspectral camera, including a band-pass filter 1, an imaging lens 2, a diaphragm 3, a relay lens 4, a dispersion element 5, and a sensor 6. The band-pass filter 1 is fixed on the imaging lens 2, the band-pass filter is placed at the forefront, and the imaging lens, diaphragm, relay lens, dispersion element, and sensor are arranged in sequence front and back on the same axis.

[0043] First, the detection object is filtered by the band-pass filter 4, and then the imaging lens 5 forms an image of the object on the plane where the aperture stop 3 is located. The aperture stop 3 is an adjustable micro-aperture stop, and its light-passing aperture is a circular hole with a light-passing diameter of 0 - 2 mm. Its main function is to limit the field of view size of the image. The image with the limited size then passes through the relay lens 4 and the dispersion element 5 in sequence and reaches the imaging plane. The function of the relay lens 4 is to focus the image at the aperture stop 3 on the sensor 6. When a beam of light passes through the dispersion element 5, several beams of light will be diffracted. The dispersion element 5 is designed with a large diffraction angle, so as to reduce the distance between the dispersion element 5 and the sensor 6. After the diffracted light reaches the sensor 6, several different images will be generated. By combining these images with the tomography principle, a two-dimensional image with spectral resolution can be reconstructed.

[0044] In a preferred embodiment, the sensor 6 is a monochromatic CCD sensor with a pixel size of 3 microns, the focal length of the imaging lens 2 is 6 mm, the focal length of the relay lens 4 is 4 mm, the imaging distance is 1254 mm, the light-passing aperture of the aperture stop 3 is set to 1 mm, the dispersion element 5 is a DOE that can generate nine images, its diameter is 2 mm, and the overall system size including the fixture is less than 3 cm × 2 cm * 2 cm.

[0045] The dispersion element 5 is one of a DOE, a two-dimensional grating or a metasurface designed with a large diffraction angle, which divides a beam of light into several different beams of light. From the center to the edge, they are the 0th order, 1st order, 2nd order... The diffraction angles of different wavelengths of light are different after diffraction. The designed dispersion element 5 with a large diffraction angle (diffractive of element, DOE) can reduce the longitudinal size of the hyperspectral camera.

[0046] The present invention also provides a hyperspectral measurement method based on the above measurement device. The measurement process includes but is not limited to the following steps:

[0047] Step 1: Before starting the measurement, it is necessary to calibrate the hyperspectral camera using a monochromator to establish the relationship between the spectrum and the image, and at the same time record the wavelength, intensity of the monochromator and the image on the sensor.

[0048] The spectral resolution of the monochromator is less than 1 nm. The light emitted by the monochromator is monochromatic point light with a wavelength range of 600 nm to 665 nm, a bandwidth of 5 nm, and a step size of 5 nm. After passing through this system, for each wavelength, the camera captures nine independent points.

[0049] Step 2: Calculate the response of the monochromatic light emitted from any point in space on the sensor according to the calibration data, and establish the correlation matrix and imaging model therein.

[0050] Step 3: After completing the calibration and model establishment, image the actual object. AsFigure 2 , which is a photograph of imaging a circular light spot in a preferred embodiment of the present invention. The middle part is the 0th order image, and the eight outer ones are the 1st order images.

[0051] The detailed description of the step 3 of establishing the model is as follows:

[0052] Suppose the size of the data cube of the object to be reconstructed after discretization is I×J×K, where I and J are the spatial dimensions and K is the spectral dimension;

[0053] For the first wavelength, the nine images formed are respectively denoted as P11_1, P12_1, P13_1, P21_1, P22_1, P23_1, P31_1, P32_1, P33_1, where P22_1 is the 0th order image of the first wavelength, and the rest are the 1st order images of the first wavelength;

[0054] After traversing all wavelengths, nine images can be obtained.

[0055] Among them, the 0th order image is the accumulation of the 0th order images of all wavelengths, that is, P22 = P22_1 + P22_2 + … + P22_K; while there is a dislocation when accumulating the 1st order images, so its final size will be slightly larger than that of the 0th order image. Taking P11 as an example, its size is [I+(K - 1)×dx]×[J+(K - 1)×dy], where dx and dy are the distances after diffraction of light of two adjacent wavelengths on the image plane respectively, and their values are determined by the calibration process in step 1. For example, if the coordinates of the image of the point light source of wavelength 1 on the plane are (x1,y1), and the coordinates of the image of the point light source of wavelength 2 on the plane are (x2,y2), then dx = x2 - x1; dy = y2 - y1.

[0056] Step 4: Reconstruct the hyperspectral image by using the MLEM-TV algorithm combined with the guided image filtering algorithm.

[0057] The iterative formula of the MLEM algorithm is:

[0058]

[0059] Where represents the reconstructed value of the jth variable after the kth iteration; A i，j represents the i-th row and j-th column of the coefficient matrix; represents the i-th projection value.

[0060] After each MLEM iteration, TV regularization is performed on the reconstructed map of each wavelength, and the purpose is to make the image of each wavelength have a reasonable gradient value.

[0061] Perform guided image filtering on the reconstructed map of each wavelength, and the purpose is to make the images of each wavelength have similar structures.

[0062] Set two termination conditions:

[0063] (1) The 2-norm of the results of two iterations is less than 10 -6 ;

[0064] (2) The number of iterations is greater than 200 times.

[0065] When one of the two conditions is satisfied, the reconstruction process is terminated, and the reconstructed value after the iteration converges is the detected hyperspectral data. Figure 3 is the image of each wavelength after reconstruction, Figure 4 is the measured spectrum.

[0066] The dynamic measurement of the hyperspectral image can be realized by repeatedly executing Step 3 to Step 4.

[0067] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field according to the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art should be within the protection scope determined by the claims.

Claims

1. A micro hyperspectral camera, characterized in that, Comprising a band-pass filter that filters out stray light outside the detection range and eliminates the influence of stray light on the measurement accuracy; an imaging lens that images the detection object at the diaphragm plane; a diaphragm that limits the size of the field of view and makes the dispersed images non-overlapping; a relay lens that focuses the image on the diaphragm onto the sensor; a dispersion element that disperses the formed image in the spectral dimension to form several different images; a sensor that acquires the dispersed image; the band-pass filter is fixed on the imaging lens, the band-pass filter is placed at the forefront, and the imaging lens, diaphragm, relay lens, dispersion element and sensor are arranged in sequence on the same axis front to back.

2. The micro hyperspectral camera according to claim 1, wherein The imaging lens and the relay lens are micro short-focus lenses, with focal lengths of 6 mm and 4 mm respectively, and their diameters are both less than 6 mm.

3. The micro hyperspectral camera according to claim 1, characterized in that, The diaphragm is an adjustable micro diaphragm, and its light passing aperture is a circular hole, and the light passing diameter is 0 - 2 mm.

4. The micro hyperspectral camera according to claim 1, wherein The dispersion element is a two-dimensional grating or a metasurface, which can divide a beam of light into several beams of different lights. From the center to the edge, they are the 0th order, 1st order, 2nd order... The diffraction angles of different wavelengths of light are different after diffraction.

5. The micro hyperspectral camera according to claim 1, wherein, The sensor is a micro monochrome CCD, with an imaging area of 2.4 mm × 2.4 mm, and the output data is 10-bit binary numbers.

6. A hyperspectral imaging method based on the micro hyperspectral camera according to any one of claims 1-5, characterized in that, Including the following steps: Step 1: Calibrate the hyperspectral camera using a monochromator, record the wavelength and intensity of the monochromator, and at the same time record the image on the sensor; Step 2: Calculate the response of the monochromatic light emitted from any point in space on the sensor according to the calibration data, and establish the correlation matrix and imaging model therein; Step 3: After completing the calibration and model establishment, image the actual object; Step 4: Reconstruct the hyperspectral image using the MLEM-TV algorithm combined with the guided image filtering algorithm, and the guided image is the 0th order image after diffraction.

7. The hyperspectral imaging method according to claim 6, characterized in that, By repeatedly executing Step 3 to Step 4, the dynamic measurement of the hyperspectral image is realized.

8. The hyperspectral imaging method according to claim 6, wherein Step 2 further includes: Step 2.1: Assume that the size of the data cube of the object to be reconstructed after discretization is I×J×K, where I and J are spatial dimensions and K is the spectral dimension; Step 2.2: For the first wavelength, the nine formed images are respectively denoted as P11_1, P12_1, P13_1, P21_1, P22_1, P23_1, P31_1, P32_1, P33_1, where P22_1 is the 0th order image of the first wavelength, and the rest are the 1st order images of the first wavelength; Step 2.3: After traversing all wavelengths, nine images are obtained.

9. The hyperspectral imaging method according to claim 6, wherein, The reconstruction algorithm in Step 4 is an iterative method, and in each iteration, MLEM reconstruction, TV regularization, and guided image filtering are performed in sequence.

10. The hyperspectral imaging method according to claim 6, wherein The spectral resolution of the monochromator is less than 1 nm.

Images