A computational optical method for obtaining translation error based on hyperspectral images

By generating a sequence of aperture point spread functions and using image deconvolution techniques to recover hyperspectral images, the problem of range and accuracy in translation error measurement in existing technologies has been solved, and large-scale, high-precision translation error extraction and sign determination have been achieved.

CN116878828BActive Publication Date: 2026-11-10INST OF OPTICS & ELECTRONICS CHINESE ACAD OF SCI
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202310882045.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-07-18
Publication Date
2026-11-10
Estimated Expiration
2043-07-18

AI Technical Summary

Technical Problem

Existing technologies cannot simultaneously achieve both wide-range and high-precision measurement of translation errors in hyperspectral image measurement, and cannot determine the sign of the translation error.

Method used

By using the parameters of the hyperspectral imaging system, a sequence of aperture spread functions is generated. The band image is then recovered using image deconvolution technology to generate an evaluation value sequence, forming a two-dimensional evaluation image. The fringe slope and centroid shift are analyzed to extract the sign and magnitude of the translation error.

Benefits of technology

It achieves high-precision translation error extraction over a wide range, accurately determines the sign of the translation error, and is unaffected by the target spectral characteristics.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116878828B_ABST
    Figure CN116878828B_ABST
Patent Text Reader

Abstract

This invention discloses a computational optical method for obtaining translation error based on hyperspectral images. The method includes: obtaining, through calibration or digital calculation, a sequence of system point spread functions (SFS) corresponding to translation errors in the range of (-λ / 2, λ / 2) for all bands using a hyperspectral image imaging device; restoring the band images using the SFS sequences corresponding to the bands using image deconvolution techniques to obtain a sequence of restored band images; obtaining an evaluation value sequence of the restored band images through a no-reference image quality assessment; repeating this process to obtain evaluation value sequences corresponding to all band images; arranging the evaluation value sequences corresponding to all band images sequentially according to band size to form a two-dimensional evaluation image; and analyzing the centroid position of the two-dimensional evaluation image along a set translation error direction or analyzing the period and slope sign along the band direction to obtain the translation error value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of optical imaging system technology, and specifically relates to a computational optical method for obtaining translation error based on hyperspectral images. Background Technology

[0002] The prior art, "Method for measuring translation error of optical multi-aperture imaging system based on hyperspectral image" (Chinese invention patent CN112432768A), discloses a method for measuring translation error of multi-aperture imaging system by using hyperspectral images of corresponding scenes or targets.

[0003] In this invention, the image quality of different bands in a hyperspectral image is evaluated using an image quality evaluation method to obtain an evaluation value for each image. Based on this evaluation value, the curve's variation period t can be calculated, and multiplying it by a coefficient C0 yields the translation error value (Piston). However, this method cannot determine the sign of the calculated translation error, and it cannot accurately measure translation errors, even for small errors (far smaller than one wavelength). Summary of the Invention

[0004] The technical problem solved by this invention is to achieve both wide-range and high-precision measurement of translation error when using hyperspectral images to measure translation error, while also being able to determine the sign of the measured translation error.

[0005] The technical solution adopted in this invention is as follows:

[0006] A computational optical method for obtaining translation error based on hyperspectral images, the method comprising the following steps:

[0007] The first step is to obtain the sequence of aperture spread functions of the hyperspectral imaging system modulated by different translation errors;

[0008] The second step is to recover the band image sequence using the point spread function sequence;

[0009] The third step is to use image quality evaluation methods to evaluate the restored image sequence of images in the same band and obtain an evaluation value sequence.

[0010] The fourth step is to arrange the evaluation value series of all band images according to the band size to form a two-dimensional evaluation image;

[0011] The fifth step is to extract the magnitude and sign of the translation error using the two-dimensional evaluation image.

[0012] Furthermore, in the first step, the parameters of the hyperspectral imaging system are known, with the shortest wavelength band being λ0 and the longest being λ. NThe translation error of the hyperspectral imaging system in the range of (-λ0 / 2, λ0 / 2) is calculated or calibrated, and the aperture spread function sequence is obtained according to a fixed step value. Store the sequence of aperture spread functions; repeat this process to obtain the sequence of aperture spread functions for all bands of the hyperspectral imaging system. in This represents the sequence of aperture spread functions for a given operating wavelength, where the translation error is between -λ0 / 2 and λ0 / 2. N This indicates that the operating wavelength is between λ0 and λ1. N Between ranges.

[0013] Furthermore, in the second step, for the band image of band λ, the corresponding aperture spread function sequence is used. By using deconvolution methods, the reconstructed image sequence corresponding to the band image is obtained. This process is repeated to obtain the restored image sequence corresponding to all band images. in This represents the reconstructed image sequence corresponding to an image in a certain operating wavelength band with a translation error between -λ0 / 2 and λ0 / 2. N This indicates that the operating wavelength is between λ0 and λ1. N Between ranges.

[0014] Furthermore, the third step involves reconstructing the image sequence from the band image of band λ. The evaluation value sequence of the restored image sequence was calculated using a reference-free image quality evaluation method. This process is repeated until the evaluation value sequence corresponding to the restored image sequence of all bands is obtained. in To reconstruct the image sequence of an image in a certain operating wavelength band with a translation error between -λ0 / 2 and λ0 / 2, (λ0, λ N This indicates that the operating wavelength is between λ0 and λ1. N Between ranges.

[0015] Furthermore, in the fourth step, the evaluation value sequence for all bands will be obtained. According to wavelength from λ0 to λ N arrangement A two-dimensional evaluation value map is generated, and the two-dimensional evaluation value map is normalized, that is, the evaluation values ​​are normalized from order 0 to 255 to form a two-dimensional evaluation image.

[0016] Furthermore, in the fifth step, the fringe slope and fringe period in the two-dimensional evaluation image are analyzed to obtain a translation error of not less than one wavelength, wherein the fringe slope determines the sign of the translation error and the fringe period determines the magnitude of the translation error; a translation error of less than one wavelength is extracted by using the offset of the centroid position of the fringe, wherein the absolute value of the offset of the centroid position determines the magnitude of the translation error and the left-right orientation of the centroid offset determines the sign of the translation error.

[0017] This invention, based on the parameters of a hyperspectral imaging system, obtains the aperture spread function sequence corresponding to all bands within the co-phase error range (-λ / 2, λ / 2) through calibration or digital calculation; using image deconvolution technology, the band images are restored using the aperture spread function sequences corresponding to the bands, resulting in a sequence of restored band images; an evaluation value sequence of the restored band images is obtained through a no-reference image quality assessment; this process is repeated to obtain the evaluation value sequence corresponding to all band images; the evaluation value sequences corresponding to all band images are arranged sequentially according to band size to form a two-dimensional evaluation image; the centroid position of the two-dimensional evaluation image is analyzed along the direction or the period and slope sign are analyzed along the band direction to obtain the translation error value.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] (1) The method of this invention achieves both large-scale and high-precision translation error extraction from hyperspectral images. Traditional methods for extracting translation errors from hyperspectral images can only directly obtain translation errors with a large range (not less than one wavelength), and it is difficult to effectively extract smaller translation errors (less than one wavelength). Therefore, the method of this invention can solve the problem that traditional methods are difficult to extract smaller translation errors, and achieve translation error extraction with both large range and high precision.

[0020] (2) The method of the present invention extracts the sign of translation error by utilizing the fringe slope or fringe centroid offset direction in the generated two-dimensional evaluation image. Traditional methods, however, cannot determine or extract the sign of translation error.

[0021] (3) The evaluation value sequence formed in the method of the present invention is generated using the same band image. Compared with the existing methods that generate an evaluation sequence using all band images, it is not sensitive to the target spectral characteristics. Therefore, the method of the present invention is not affected by the target spectral curve. Attached Figure Description

[0022] Figure 1 This is the process flow of an embodiment of the method of the present invention.

[0023] Figure 2 This is a step in this embodiment of the invention to obtain a restored image sequence of a band image through a point spread function sequence.

[0024] Figure 3 This is the step in this embodiment of the invention to obtain the evaluation value sequence of the restored image sequence.

[0025] Figure 4 This is an example of a method for calculating the absolute value of translation error when the measured translation error is large (not less than half a wavelength) according to an embodiment of the present invention.

[0026] Figure 5 This is an example of a method for determining the sign of translation error when the measured translation error is large (not less than half a wavelength) according to an embodiment of the present invention.

[0027] Figure 6 This is an example of a method for calculating translation error when the measured translation error is small (less than half a wavelength) according to an embodiment of the present invention. Detailed Implementation

[0028] The method of the present invention will be further described below with reference to the accompanying drawings and embodiments, but this should not be construed as limiting the scope of protection of the present invention.

[0029] Figure 1 These are the steps of an embodiment of the method of the present invention, such as... Figure 1 As shown, the method for obtaining translation error over a wide range with high precision based on hyperspectral images mainly includes the following steps: Step ①: Obtaining a reconstructed image sequence of the band image through a point spread function sequence; Step ②: Obtaining an evaluation value sequence of the reconstructed image sequence; Step ③: Generating a two-dimensional evaluation image from the evaluation value sequence; and Step ④: Calculating the translation error (piston) from the two-dimensional evaluation image. In this embodiment, the band ranges from 501 nm to 600 nm, with a spectral resolution of 1 nm.

[0030] like Figure 2 As shown, the present invention describes the steps for obtaining a reconstructed image sequence of a band image through a point spread function sequence. For example, for a 501nm band image, firstly, the point spread function of the 501nm band imaging system with a translation error between -250nm and 250nm is obtained in 25nm increments, forming a point spread function sequence; then, the 501nm band image and the point spread functions in the point spread function sequence are sequentially deconvolved to obtain the reconstructed image, forming a reconstructed image sequence of the band image.

[0031] like Figure 3 As shown, the method for obtaining the evaluation value sequence of the restored image sequence in this embodiment of the invention is as follows: For example, the evaluation value of each image in the restored image sequence of the 501nm band image is calculated using the same method and the set no-reference evaluation algorithm to form an evaluation value sequence (4), and an evaluation image (5) is formed based on the evaluation value sequence.

[0032] like Figure 1As shown, in step ③, the evaluation image is formed from the evaluation value sequence. The evaluation image of the band image is arranged in order of band size to form a two-dimensional evaluation image. One direction of the two-dimensional evaluation image is the band direction, and the other direction is the direction of change of translation error.

[0033] Figure 1 The embodiment of step ④ in the process includes Figure 4 , Figure 5 and Figure 6 As shown.

[0034] Figure 4 Example: When the measured translation error is large (greater than half a wavelength), the absolute value of the translation error is calculated. Figure 4 The example calculates the magnitude of the stripe interval period P in a two-dimensional evaluation image, and then calculates the translation error based on P. For example... Figure 4 As shown, the corresponding fringe spacing periods are P1 and P2, respectively, when the translation errors are -20 μm and -60 μm. The absolute value of the translation error can be calculated by inversely proportional to the spacing period and the absolute value of the translation error.

[0035] pass Figure 5 As shown, the sign of the translation error, i.e., its positive or negative nature, can be determined based on the slope of the stripes in the two-dimensional evaluation image. Figure 5 As shown in the example, for two-dimensional evaluation images with translation errors of 60μm and -60μm, the fringe slopes are k1 and k2, respectively. With the band direction as the reference, the signs of k1 and k2 are different from those of the slope. Therefore, the sign of the translation error can be determined by the sign of the slope.

[0036] When the measured translation error is small (less than half a wavelength), it can be achieved through... Figure 6 The example calculates the translation error by determining the position of the fringe centroids in a two-dimensional evaluation image. Three two-dimensional evaluation images with translation errors of -0.05μm, 0μm, and 0.05μm are used as examples. The fringe centroids in each band of the evaluation image are shown by the dashed lines. The spacing between the fringe centroids in the 0.05μm and 0μm two-dimensional evaluation images is μ0, and the spacing between the fringe centroids in the -0.05μm and 0μm two-dimensional evaluation images is -μ0. Here, the magnitude of μ0 is proportional to the absolute value of the translation error, and the sign before μ0 determines the sign of the translation error.

[0037] Using the above method, translation error can be extracted over a wide range and with high precision from hyperspectral images containing translation error information.

Claims

1. A computational optical method for obtaining translation error based on hyperspectral images, characterized in that, The method includes the following steps: The first step is to obtain the sequence of aperture spread functions of the hyperspectral imaging system modulated by different translation errors; The second step is to recover the band image sequence using the aperture spread function sequence; The third step is to use image quality evaluation methods to evaluate the restored image sequence of images in the same band and obtain an evaluation value sequence. The fourth step is to arrange the evaluation value series of all band images according to the band size to form a two-dimensional evaluation image; The fifth step involves extracting the magnitude and sign of the translation error from the two-dimensional evaluation image. This includes: analyzing the fringe slope and fringe period within the two-dimensional evaluation image to obtain a translation error of not less than one wavelength, where the fringe slope determines the sign of the translation error and the fringe period determines the magnitude of the translation error; and extracting a translation error of less than one wavelength from the offset of the centroid position of the fringe, where the absolute value of the centroid position offset determines the magnitude of the translation error and the left-right orientation of the centroid offset determines the sign of the translation error.

2. The computational optical method for obtaining translation error based on hyperspectral images according to claim 1, characterized in that, In the first step, the parameters of the hyperspectral imaging system are known, with the shortest operating wavelength being λ0 and the longest operating wavelength being... The translation error of the hyperspectral imaging system in the range of (-λ0 / 2, λ0 / 2) is obtained through digital calculation or calibration, and the aperture spread function sequence is obtained according to a fixed step value. Store the aperture spread function sequence; repeat this process to obtain aperture spread function sequences for all bands of the hyperspectral imaging system. ;in This indicates that at a certain operating wavelength, the translation error is within... arrive The sequence of aperture spread functions for that wavelength. Indicates the operating wavelength at arrive Between ranges.

3. The computational optical method for obtaining translation error based on hyperspectral images according to claim 2, characterized in that, In the second step, for the band image of band λ, the corresponding aperture spread function sequence is used. By using deconvolution methods, the reconstructed image sequence corresponding to the band image is obtained. Repeat this process to obtain the restored image sequence corresponding to all band images. ;in, This indicates that at a certain operating wavelength, the translation error is within... arrive The reconstructed image sequence corresponding to the band images between them. Indicates the operating wavelength at arrive Between ranges.

4. The computational optical method for obtaining translation error based on hyperspectral images according to claim 3, characterized in that, The third step involves restoring the image sequence from the band image of band λ. The evaluation value sequence of the restored image sequence was calculated using a reference-free image quality evaluation method. This process is repeated until the evaluation value sequence corresponding to the restored image sequence of all bands is obtained. ;in, To ensure that the translation error is within a certain operating wavelength arrive Image sequence recovery from interband images, Indicates the operating wavelength at arrive Between ranges.

5. The computational optical method for obtaining translation error based on hyperspectral images according to claim 4, characterized in that, In the fourth step, the evaluation value sequence for all bands will be obtained. According to wavelength arrive arrangement A two-dimensional evaluation value map is generated, and the two-dimensional evaluation value map is normalized, that is, the evaluation values ​​are normalized from order 0 to 255 to form a two-dimensional evaluation image.

Citation Information

Patent Citations

  • Optical multi-aperture imaging system translation error measurement method based on hyperspectral image

    CN112432768A

  • Closed-loop correction control method for phase translation error of spliced telescope system

    CN108519671A

  • Spliced mirror co-phase error correction method based on dispersion fringe slope analysis

    CN111207910A