Two-dimensional spectral demodulation method in fiber optic spectral confocal displacement sensing system

Abstract

The present invention discloses a two-dimensional spectrum demodulation method in a fiber optic spectral confocal displacement sensing system, which belongs to the field of optical precision measurement. In order to solve the problem of low spectral resolution of existing spectral demodulation methods, the method uses the normalized confocal axial response characteristics in the pre-calibration stage to determine the corresponding position relationship of pixels of different diffraction order spots of the same wavelength in the two-dimensional spectrum diagram, and then determines the boundary points of the free spectrum region by using the characteristic that the light intensity of corresponding pixels of the same wavelength in adjacent diffraction order spectra is the same; in the measurement stage, the pixel light intensity in the free spectrum region of the two-dimensional spectrum diagram is compensated and restored to a one-dimensional spectrum using the results obtained during pre-calibration, and the position value of the measured surface is obtained by using the quantitative one-to-one correspondence between the peak wavelength of the spectrum and the position of the measured surface. The present invention effectively utilizes the two-dimensional spectrum diagram information, improves the spectral resolution of the system, and can effectively improve the measurement resolution of the fiber optic spectral confocal displacement sensor.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical precision measurement, and in particular relates to a two-dimensional spectrum demodulation method in an optical fiber spectrum confocal displacement sensing system. Background Art

[0002] Spectral confocal displacement sensing technology uses the principle of optical dispersion to converge light beams of different wavelengths at different optical axis positions. After being reflected by the measured surface, the light that meets the confocal condition is received by the spectral demodulation module, and the peak of the spectrum is extracted. The one-to-one correspondence between the spectral peak wavelength and the position of the measured surface is used to achieve displacement measurement. This technology has the advantages of a large measurement range and high precision, and is widely used in micro-optical and micro-mechanical parts processing, aerospace, and medical equipment. Among them, the miniaturization of the measuring probe is one of the important research directions in spectral confocal displacement sensing technology. It is of great significance for the precise measurement of component characteristic parameters in cases where the measuring space is narrow and the installation of the measuring probe is restricted. The measuring probe of the fiber optic spectral confocal displacement sensing system is small, but the miniaturization of the probe makes it difficult to further improve the displacement resolution of the system, which seriously limits the application expansion of this type of sensor.

[0003] Since system resolution is determined by both the probe's dispersion coefficient and the spectral resolution of the spectral demodulation module, improving the spectral resolution of the spectral demodulation module can improve the axial displacement resolution of the fiber spectral confocal system. Currently, spectral demodulation modules primarily employ a one-dimensional dispersive structure, where a wide-spectrum beam is dispersed in a single direction, forming a line spectrum on a CCD / CMOS detector. Improving the spectral resolution of this structure requires expanding the spectral dispersion region and increasing the dispersion interval between adjacent wavelengths. However, limitations in the focusing field of view, CCD pixel size, and array size make further improvements in spectral resolution difficult. To address this issue, the paper "Research on the Design and Calibration Method of a Broadband Echelle Grating Spectrometer" proposes a cross-dispersion structure that achieves both wideband and high-resolution spectral demodulation. This structure uses an echelle grating as the primary dispersive element, combined with a low-dispersion element to cross-disperse the spectrum, forming a two-dimensional spectral image on the array detector. This structure folds the signal spectrum into a two-dimensional space, resulting in a much larger wavelength dispersion region than a one-dimensional dispersive structure. This allows for higher spectral resolution within the same signal spectral range. Therefore, the cross-dispersion structure is more suitable for improving the axial resolution of the fiber spectral confocal displacement sensor system with a miniaturized dispersion probe.

[0004] For the cross-dispersion structure, the accuracy of its spectrum restoration will affect the accuracy of the peak wavelength extraction of the spectrum signal. Figure 1The process of obtaining a one-dimensional spectrum (i.e., a line spectrum) from a two-dimensional spectrum begins by determining the free spectral region in the two-dimensional spectrum, which contains the primary energy of the diffracted light. The wavelengths of each band within this region are then connected end-to-end to establish a one-to-one correspondence between wavelength and pixel position. The paper "High-accuracy spectral reduction algorithm for the echelle spectrometer" (Applied Optics, 2016) establishes a mathematical model between pixel position and wavelength based on geometric optics. Using the optical parameters of the cross-dispersion structure as variables, the relationship between characteristic spectral lines and pixel position is obtained through a wavelength calibration process. The optical parameters are then modified to obtain the final spectral reduction model. Based on this model, the intensity of each wavelength is extracted from the corresponding pixel position, resulting in a one-dimensional spectrum. The paper "New spectral reduction algorithm for echelle spectrometer in laser-induced breakdown spectroscopy" (Optics Express, 2018) models the spectrum, describing the relationship between wavelength and pixel position using a polynomial. Using known characteristic spectral lines and pixel positions, the coefficients of the polynomial are iteratively calculated to obtain a mapping between wavelength and pixel position.

[0005] However, existing spectral restoration methods have the following problems: On the one hand, due to factors such as the nonlinear dispersion of the prism, the aberration of the optical system, and the installation deviation of each optical component, there is a certain difference between the constructed spectral restoration model and the actual distribution of the spectrum; on the other hand, when calibrating the wavelength, the characteristic spectral lines used are limited. The current method can only accurately correspond the characteristic spectral lines to the pixel positions, but due to the nonlinearity of the dispersion, it is difficult to ensure the accuracy of the relationship between other wavelengths and pixel positions. In addition, the existing spectral restoration process only uses the spectral part within the free spectral region, and splices the in-region parts of each level of spectrum end to end to obtain a one-dimensional spectrum, but the diffraction efficiency of each wavelength is not consistent, such as Figure 2 As shown, this seriously interferes with the spectral peak wavelength detection of the fiber spectral confocal displacement sensing system, thereby reducing the accuracy of the system displacement measurement. Summary of the Invention

[0006] In order to overcome the shortcomings of the existing technology, the present invention provides a two-dimensional spectral demodulation method in a fiber optic spectral confocal displacement sensing system. This method can not only improve the accuracy of wavelength and pixel position mapping in spectral restoration, but also reduce the uneven diffraction efficiency of each sampling wavelength caused by the cross-dispersion structure, thereby reducing the peak wavelength positioning error of the spectral signal and improving the measurement accuracy of the displacement fiber optic spectral confocal displacement sensing system.

[0007] In order to achieve the above object, the purpose of the present invention is to be achieved through the following technical solutions:

[0008] The present invention demodulates the spectral signal of the fiber spectral confocal displacement sensing system through a cross-dispersion structure. In the system, a light source emits a wide-spectrum light beam, which is transmitted from port 1 to port 2 of a fiber coupler, enters a micro-dispersion probe, and then irradiates a measured surface. The reflected signal light is sequentially transmitted through the micro-dispersion probe, port 2, and port 3 of the fiber coupler to the cross-dispersion structure. After passing through a collimator, the signal light beam generates primary dispersion along the y-axis direction through an echelon grating, wherein the sub-spectra overlap with each other. Subsequently, the signal light beam is secondary dispersed by a dispersion prism, and the overlapping spectra are dispersed along the x-direction. The overlapping spectra are then converged by a focusing lens to form an initial image on the detection plane of a CCD array detector. After downsampling and spectral restoration processing, the displacement of the measured surface can be measured by utilizing the one-to-one correspondence between the spectral peak wavelength and the position of the measured surface.

[0009] In order to achieve spectral restoration, that is, to obtain a one-dimensional spectrum from a two-dimensional spectral image, the present invention provides a two-dimensional spectrum demodulation method in a fiber spectral confocal displacement sensing system, comprising the following steps:

[0010] S1: During pre-calibration, the normalized confocal axial response characteristic curve of each sampling wavelength (i.e., each pixel on the two-dimensional spectrum) is obtained through the fiber spectral confocal system;

[0011] S2: determining the corresponding positional relationship between different diffraction order spots of the same wavelength and pixels in the two-dimensional spectrum graph through the normalized confocal axial response characteristic curve;

[0012] S3: During pre-calibration, the light from the white light source is coupled into the fiber coupler, and then a two-dimensional spectrum is obtained after passing through the cross-dispersion structure. The boundary points of the free spectrum region in the two-dimensional spectrum are then obtained by the diffraction intensity.

[0013] S4: During measurement, using the results of step S2 and step S3, the pixel intensities outside the free spectrum region in the two-dimensional spectrum graph are compensated to the pixels corresponding to the same wavelength within the free spectrum region;

[0014] S5: restoring the spectrum of the compensated free spectrum region to a one-dimensional spectrum, and extracting the peak wavelength of the spectrum.

[0015] Furthermore, steps S1 to S3 are a pre-calibration process, which is completed when the optical fiber spectral confocal displacement sensing system is manufactured; steps S4 and S5 are the process of demodulating the spectral signal when the system is used for measurement.

[0016] Furthermore, the method for obtaining the two-dimensional spectrum in the above steps is: downsampling the initial image I0(x,y) obtained by the array detector (405) to generate a two-dimensional spectrum I(x,y), specifically:

[0017]

[0018] In the above formula, E is the set of adjacent pixels in the same row (y=y0) in the x direction of a certain diffraction order spectrum in the image I0(x,y), and y0 represents any row in the order spectrum.

[0019] Because the three-port core of the fiber coupler has a certain size, after cross-dispersion, light of the same wavelength in the two-dimensional spectrum is distributed in adjacent diffraction spectra. The dispersion interval of the echelle grating is large, while the dispersion interval of the prism is small. In the spectrum, light of the same order and wavelength is approximately distributed in several adjacent pixels in the x-direction, while light of different wavelengths of the same order is distributed in different rows. Therefore, by downsampling the initial image to obtain a two-dimensional spectrum, the diffraction spot of each wavelength order is approximately represented by a single pixel.

[0020] Furthermore, step S1 is specifically as follows: the measured surface is moved along the optical axis direction of the micro-dispersion probe with a step length Δz, and the initial image corresponding to each axial position z is recorded; for each initial image at position z of the measured surface, a two-dimensional spectrum is obtained by downsampling; in the two-dimensional spectrum, for each pixel position (x, y), the change in pixel brightness at different z positions can be used to obtain the normalized confocal axial response characteristic curve I of the corresponding pixel position. z (x,y).

[0021] Furthermore, step S2 is specifically as follows: in the two-dimensional spectrum graph I(x, y), for two pixels in different diffraction order spectrum bands, their normalized cross-correlation coefficient (NCC) is calculated according to formula (2). When the NCC value is the largest, the matching degree of the two confocal axial response characteristic curves is the highest, that is, the two pixels represent different positions corresponding to the same wavelength of light in the two-dimensional spectrum graph:

[0022]

[0023] In the above formula, I z (x1,y1) and I z (x2, y2) are the normalized axial response characteristic curves of two pixels in different diffraction order spectral bands, and Z is the set of different axial positions z within the displacement measurement range.

[0024] Furthermore, step S3 is used to determine the boundary of the free spectrum region on the two-dimensional spectrum graph, specifically: directly coupling a white light source into port 2 of the fiber coupler, and then emitting light from port 3 of the fiber coupler, after passing through a cross-dispersion structure, to obtain an initial image, and then downsampling to obtain a two-dimensional spectrum graph; when the end pixel of the m+1 level spectrum and the beginning pixel of the m level spectrum simultaneously meet the following two conditions, then the pixel pair is the boundary point of the free spectrum region;

[0025] The two conditions are: condition 1) the normalized axial characteristic curves of the two pixels are the same or similar, that is, the pixels are of the same wavelength; condition 2) the diffraction intensity at the end of the m+1 level spectrum and the beginning of the m level spectrum are substantially the same;

[0026] Wherein, the light power distribution of each wavelength of the white light source is approximately uniform within the measurement spectrum range.

[0027] When white light enters the cross-dispersion structure, the free spectrum region of the two-dimensional spectrum diagram has the following characteristics: 1) the wavelengths of adjacent order spectra are connected end to end in the free spectrum region; 2) the intensity of the same sampling wavelength in the free spectrum region is greater than its intensity outside the region; 3) the spectra of different orders are clearly separated in the spectrum diagram.

[0028] Furthermore, step S4 is used to equalize the diffraction efficiency of light at each wavelength to facilitate accurate restoration of the one-dimensional spectrum. Specifically, during measurement, when the measured surface is at a certain position z, an initial image is obtained using the fiber spectral confocal displacement sensing system, and a two-dimensional spectrum is obtained by downsampling. Based on the correspondence between wavelength and pixel position obtained in step S2 during pre-calibration and the position of the boundary points of the free spectrum region obtained in step S3, the two-dimensional spectrum is processed. Specifically, the pixel intensity outside the free spectrum region is compensated to the pixel corresponding to the same wavelength within the free spectrum region, thereby equalizing the diffraction efficiency of light at each wavelength, thereby reducing the impact of inconsistent diffraction efficiency on the extraction of the spectral peak wavelength.

[0029] Furthermore, the step S5 is used to accurately restore the one-dimensional spectrum and extract the peak wavelength of the spectrum, and obtain the position value of the measured surface by using the quantitative one-to-one correspondence between the spectrum peak wavelength and the position of the measured surface.

[0030] The present invention has the following significant features and beneficial effects:

[0031] 1. In the present invention, the intensity of each sampling wavelength is obtained by summing its intensities inside and outside the free spectrum region. This approach reduces the intensity differences between the sampling wavelengths in the free spectrum region caused by the different diffraction efficiencies of the dispersion device during spectral restoration, improves the accuracy of locating the peak wavelength of the one-dimensional spectrum, and thus improves the measurement accuracy of the displacement.

[0032] 2. The present invention utilizes the different confocal axial response characteristic curves of light of different wavelengths in the fiber spectral confocal displacement sensing system. Using a normalized cross-correlation algorithm, it accurately matches different pixel positions in the spectrum for light of different orders of the same wavelength. Compared to traditional two-dimensional spectral restoration methods, the present invention not only more accurately divides the free spectral region but also provides an accurate pixel correspondence for compensating the intensity within the free spectral region, thereby obtaining a more accurate one-dimensional spectrum. This invention eliminates the need for a complex spectral restoration model or the use of characteristic spectral lines for calibration, making it simple and easy to implement. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a two-dimensional spectrum image of white light formed by the cross-dispersion structure;

[0034] Figure 2 is the relative intensity curve of each wavelength in the free spectrum region of the echelle grating;

[0035] Figure 3 Schematic diagram of the optical fiber spectral confocal displacement sensing system combined with a cross-dispersion structure;

[0036] Figure 4 A schematic flow chart of the two-dimensional spectrum demodulation method provided by the present invention;

[0037] Figure 5 Schematic diagram of two-dimensional spectrum distribution on the array detector;

[0038] Figure 6 This is a schematic diagram of the matching of different pixel positions corresponding to the same sampling wavelength in the two spectra;

[0039] Figure 7 Schematic diagram of the optical path used to determine the free spectrum range during calibration;

[0040] Figure 8 Schematic diagram of the boundary of the free spectral region in the two-dimensional spectrum diagram;

[0041] In the figure, 10, light source module; 20, fiber coupler; 30, miniature dispersion probe; 40, cross-dispersion structure; 401, collimator; 402, echelle grating; 403, dispersion prism; 404, collimator; 405, area array detector; 50, surface of the object to be measured. DETAILED DESCRIPTION

[0042] The following are specific embodiments of the present invention and the accompanying drawings to describe the technical solutions of the present invention in detail, but the present invention is not limited to these embodiments.

[0043] like Figure 3As shown, the present invention demodulates the spectrum signal of the optical fiber spectral confocal displacement sensing system through a cross-dispersion structure (40). In the system, a light source (10) emits a wide spectrum light beam, which is transmitted to port 2 through port 1 of an optical fiber coupler (20), enters a micro-dispersion probe (30), and then irradiates a measured surface (50). The reflected signal light is sequentially transmitted through the micro-dispersion probe (30), port 2, and port 3 of the optical fiber coupler (20) to the cross-dispersion structure (40). After passing through a collimator (401), the signal light beam passes through an echelle grating (402) and generates a primary dispersion along the y-axis direction, wherein the sub-spectra overlap with each other. Subsequently, the signal light beam is secondary dispersed through a dispersion prism (403), and the overlapping spectra are dispersed along the x-direction. The overlapping spectra are then converged through a focusing lens (404) to form an initial image on the detection plane of a planar array detector CCD (405). After downsampling and spectrum restoration processing, the displacement of the measured surface can be measured by utilizing the one-to-one correspondence between the spectrum peak wavelength and the position of the measured surface.

[0044] In order to restore a one-dimensional spectrum from an image detected by a planar array detector (405), the present invention proposes a two-dimensional spectrum demodulation method in a fiber optic spectrum confocal displacement sensing system, such as Figure 4 The specific steps are as follows:

[0045] S1: During pre-calibration, the normalized confocal axial response characteristic curve of each sampling wavelength (i.e., each pixel on the two-dimensional spectrum) is obtained through the fiber spectral confocal system;

[0046] S2: determining the corresponding positional relationship between different diffraction order spots of the same wavelength and pixels in the two-dimensional spectrum graph through the normalized confocal axial response characteristic curve;

[0047] S3: During pre-calibration, light from the white light source (10) is coupled into the optical fiber coupler (20), and then passes through the cross-dispersion structure (40) to obtain a two-dimensional spectrum, and then the boundary points of the free spectrum region in the two-dimensional spectrum are obtained by diffraction intensity;

[0048] S4: During measurement, using the results of step S2 and step S3, the pixel intensities outside the free spectrum region in the two-dimensional spectrum graph are compensated to the pixels corresponding to the same wavelength within the free spectrum region;

[0049] S5: restoring the spectrum of the compensated free spectrum region to a one-dimensional spectrum, and extracting the peak wavelength of the spectrum.

[0050] Furthermore, steps S1 to S3 are a pre-calibration process, which is completed when the optical fiber spectral confocal displacement sensing system is manufactured; steps S4 and S5 are a process of demodulating the spectral signal when the system is used for measurement.

[0051] Furthermore, the method for obtaining the two-dimensional spectrum image in the above steps is: down-sampling the initial image I0(x,y) obtained by the array detector (405) according to formula (1) to generate a two-dimensional spectrum image I(x,y).

[0052] The initial image to be demodulated is distributed as follows Figure 5 As shown, it contains multiple order spectra, and the diffraction order decreases from left to right. The coordinate system is constructed with the pixel in the lower left corner of the image as the origin. The x-axis corresponds to the prism dispersion direction; the y-axis corresponds to the echelle dispersion direction. The positive directions of the x-axis and y-axis are the directions of increasing wavelength.

[0053] Furthermore, the step S1 is specifically as follows: the measured surface is moved along the optical axis direction of the micro-dispersion probe (30) with a step length Δz, and the initial image corresponding to each axial position z is recorded; for the initial image at each measured surface position z, a two-dimensional spectrum graph I (x, y) is obtained by downsampling; in the two-dimensional spectrum graph, for each pixel position (x, y), the change of pixel brightness at different z positions is used to obtain the normalized axial confocal response characteristic curve I of the corresponding pixel position z (x,y).

[0054] Furthermore, step S2 is specifically as follows: in the two-dimensional spectrum graph I(x, y), for two pixels of different diffraction orders, their normalized cross-correlation coefficient (NCC) is calculated according to formula (2). When the NCC value is the largest, the matching degree of the two axial characteristic curves is the highest, that is, the two pixels represent different positions corresponding to the same wavelength of light in the two-dimensional spectrum graph:

[0055] like Figure 6 As shown, the position of a pixel (x m+1 ,y m+1 ) are respectively related to the m-level spectrum (x m ,y m )、(x' m ,y' m )、(x” m ,y” m ) and the normalized confocal axial response characteristic curves of the pixel positions are compared. If (x m+1 ,y m+1 ) and (x m ,y m ) is the best match, then the two pixel positions correspond to the same sampling wavelength.

[0056] Furthermore, the step S3 is used to determine the boundary of the free spectrum region on the two-dimensional spectrum graph, specifically: Figure 7As shown, the white light source (10) is directly coupled into the 2nd port of the optical fiber coupler (20), and then emitted from the 3rd port of the optical fiber coupler (20). After passing through the cross dispersion structure (40), an initial image is obtained; and then a two-dimensional spectrum is obtained by downsampling, as shown in FIG. Figure 8 As shown in , when the pixel at the end of the m+1-level spectrum and the pixel at the beginning of the m-level spectrum simultaneously meet the following two conditions, the pixel pair is the boundary point of the free spectrum region;

[0057] Condition 1) The normalized axial characteristic curves of the two pixels are identical or similar, that is, the pixels have the same wavelength; Condition 2) The diffraction intensities of the two pixels at the end of the m+1-order spectrum and the beginning of the m-order spectrum are basically the same;

[0058] Wherein, the optical power distribution of each wavelength of the white light source is approximately uniform within the measurement spectrum range.

[0059] When white light enters the cross-dispersion structure, the free spectrum region of the two-dimensional spectrum diagram has the following characteristics: 1) the wavelengths of adjacent order spectra are connected end to end in the free spectrum region; 2) the intensity of the same sampling wavelength in the free spectrum region is greater than its intensity outside the region; 3) the spectra of different orders are clearly separated in the spectrum diagram.

[0060] Furthermore, the step S4 is used to balance the diffraction efficiency of each wavelength of light, so as to accurately restore the one-dimensional spectrum. Specifically, when measuring, Figure 3 As shown, when the measured surface (50) is at a certain position z, an initial image is obtained by the optical fiber spectral confocal displacement sensing system, and a two-dimensional spectrum diagram is obtained by downsampling; based on the correspondence between the wavelength and the pixel position obtained in step S2 during pre-calibration and the boundary point position of the free spectrum region obtained in step S3, the two-dimensional spectrum diagram is processed, that is, the pixel intensity outside the free spectrum region is compensated to the pixel corresponding to the same wavelength in the free spectrum region, so as to balance the diffraction efficiency of light of each wavelength, thereby reducing the influence of inconsistent diffraction efficiency on the extraction of spectral peak wavelength.

[0061] Since the diffracted light energy is mainly concentrated in the free spectrum region, and the diffraction efficiency of each wavelength is not consistent, in order to balance the diffraction efficiency of each wavelength in the free spectrum region, the diffracted light energy outside the free spectrum region is compensated to the free spectrum region to reduce the impact of the inconsistent diffraction efficiency on the peak wavelength extraction.

[0062] Furthermore, step S5 is used to accurately restore the one-dimensional spectrum and extract the peak wavelength of the spectrum, and then use the quantitative one-to-one correspondence between the spectrum peak wavelength and the position of the measured surface to obtain the position value z of the measured surface.

[0063] The above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, but these descriptions should not be understood as limiting the scope of the present invention.

[0064] The protection scope of the present invention is defined by the appended claims, and any modifications based on the claims of the present invention are within the protection scope of the present invention.

Claims

1. A two-dimensional spectrum demodulation method in a fiber spectral confocal displacement sensing system, characterized in that: The following steps are involved: S1: During pre-calibration, the normalized confocal axial response characteristic curve of each sampling wavelength is obtained through the fiber spectral confocal displacement sensing system; S2: determining the corresponding positional relationship between different diffraction order spots of the same wavelength and pixels in the two-dimensional spectrum graph using the normalized confocal axial response characteristic curve; S3: During pre-calibration, the light from the white light source is coupled into the fiber coupler, and then a two-dimensional spectrum is obtained after passing through the cross-dispersion structure. The boundary points of the free spectrum region in the two-dimensional spectrum are then obtained by the diffraction intensity. S4: During measurement, using the results of step S2 and step S3, the pixel intensities outside the free spectrum region in the two-dimensional spectrum graph are compensated to the pixels corresponding to the same wavelength within the free spectrum region; S5: restoring the spectrum of the compensated free spectrum region to a one-dimensional spectrum, and extracting the peak wavelength of the spectrum; Among them, steps S1 to S3 are the pre-calibration process, which is completed when the fiber spectral confocal displacement sensing system is manufactured; steps S4 and S5 are the process of demodulating the spectral signal when using the system for measurement; Wherein, the optical fiber spectral confocal displacement sensing system comprises: A broad spectrum light beam emitted by a light source (10) is coupled into port 1 of a fiber coupler (20), transmitted through port 2 to a micro-dispersion probe (30), and then irradiated onto a surface to be measured (50). The reflected light is sequentially transmitted through the micro-dispersion probe (30), ports 2 and 3 of the fiber coupler (20) to a cross-dispersion structure (40). After passing through a collimator (401), the light is dispersed in the y direction by an echelle grating (402), and then secondarily dispersed in the x direction by a dispersion prism (403). After passing through a focusing lens (404), an initial image is obtained on a detection plane of a planar array detector (405). The method for obtaining the two-dimensional spectrum in the step is as follows: the initial image obtained by the array detector (405) I 0(x, y) is downsampled to produce a two-dimensional spectrum I (x, y), specifically: In the above formula, E For images I The set of adjacent pixels in the same row (y = y0) in the x direction of a certain diffraction order spectrum in 0(x, y), where y0 represents any row in the order spectrum; The step S1 includes: The measured surface is moved along the optical axis direction of the micro-dispersion probe (30) with a step length Δz, and the initial image corresponding to each axial position z is recorded; for the initial image at each position z of the measured surface, a two-dimensional spectrum is obtained by downsampling; in the two-dimensional spectrum, for each pixel position (x, y), the normalized confocal axial response characteristic curve of the corresponding pixel position is obtained by using the pixel brightness change at different z positions. I z (x, y); The step S2 includes: In the two-dimensional spectrum I In (x, y), for two pixels in different diffraction order spectral bands, their normalized cross-correlation coefficient (NCC) is calculated according to the following formula. When the NCC value is the largest, the matching degree of the two axial characteristic curves is the highest, that is, the two pixels represent the same wavelength light at different positions in the two-dimensional spectrum. In the above formula, I z (x1, y1) and I z (x2, y2) are the normalized confocal axial response characteristic curves of two pixels in different diffraction order spectral bands, and Z is the set of different axial positions z within the displacement measurement range.

2. The two-dimensional spectrum demodulation method in a fiber spectral confocal displacement sensing system according to claim 1, characterized in that: The step S3 includes: A white light source is directly coupled into port 2 of an optical fiber coupler (20), and then emitted from port 3 of the optical fiber coupler (20). After passing through a cross-dispersion structure (40), an initial image is obtained, and then a two-dimensional spectrum is obtained by downsampling. When the end pixel of the m+1 level spectrum and the beginning pixel of the m level spectrum simultaneously meet the following two conditions, the pixel pair is a boundary point of the free spectrum region; The two conditions are: 1) the normalized confocal axial response characteristic curves of the two pixels are identical or similar, that is, the pixels are of the same wavelength; 2) the diffraction intensity at the end of the m+1 order spectrum and the beginning of the m order spectrum are substantially the same; Wherein, the light power distribution of each wavelength of the white light source is approximately uniform within the measurement spectrum range.

3. The two-dimensional spectrum demodulation method in a fiber spectral confocal displacement sensing system according to claim 1, characterized in that: The step S4 includes: During measurement, when the measured surface is at a certain position z, an initial image is obtained through the fiber optic spectral confocal displacement sensing system, and a two-dimensional spectrum diagram is obtained by downsampling. Based on the correspondence between wavelength and pixel position obtained in step S2 during pre-calibration and the boundary point position of the free spectrum area obtained in step S3, the two-dimensional spectrum diagram is processed, that is, the pixel intensity outside the free spectrum area is compensated to the pixel corresponding to the same wavelength within the free spectrum area, so as to balance the diffraction efficiency of light of each wavelength.

Images