A high-precision phase unwrapping method based on fringe order correction

Abstract

The present application belongs to the field of visual detection, and provides a high-precision phase unwrapping method based on fringe order correction, which is used to solve the problems existing in the current time phase unwrapping method and the space phase unwrapping method; the present application firstly processes the collected image of the camera to obtain a single-frequency, high-frequency wrapped phase, and simultaneously completes the identification of the defect area; then the space phase unwrapping method is used to unwrap the high-frequency wrapped phase, and the fringe order is inversely calculated; the single-frequency wrapped phase is expanded to calculate the reference order, and the fringe order is corrected in the defect area; finally, the order difference between the reference phase and the space phase unwrapping result is calculated, the corrected fringe order is superimposed as the real order of the defect area, the fringe order is taken as the real order of the non-defect area, and the phase unwrapping result is calculated according to the real order; the present application improves the robustness of the phase unwrapping to the defect area through the correction of the fringe order, and high-precision unwraps the high-frequency fringe without increasing the number of projection frames.

Description

Technical Field

[0001] This invention belongs to the field of visual inspection, specifically providing a high-precision phase unfolding method based on fringe order correction. Background Technology

[0002] In the field of phase-based 3D measurement, phase unwrapping technology, as a core component, has long received widespread attention from academia and industry. During phase unwrapping, continuous absolute phase values ​​are recovered by analyzing phase transitions in the wrapped phase map, providing accurate phase data support for subsequent 3D reconstruction. The accuracy of phase unwrapping directly determines the final accuracy of 3D measurement, while its computational efficiency directly affects the real-time performance of the entire 3D reconstruction system. In applications such as industrial inspection, biomedical imaging, and digital preservation of cultural heritage, increasingly higher demands are placed on the accuracy and speed of 3D reconstruction. Therefore, developing efficient and robust phase unwrapping methods to achieve high-precision, high-speed 3D reconstruction has become an important research topic in the field of 3D measurement.

[0003] As the two most commonly used phase unfolding methods, temporal phase unfolding and spatial phase unfolding each have their own advantages and disadvantages. Temporal phase unfolding has advantages such as high unfolding accuracy and error not propagating along the unfolding path, but it often requires projecting multiple sets of fringes of different frequencies, and multi-frequency phase unfolding methods risk amplifying errors. Spatial phase unfolding only requires fringes of a single frequency, making it suitable for unfolding smooth regions, but it is limited by the assumption of surface smoothness, and errors will propagate along the unfolding path. To overcome the shortcomings of both methods and strive to ensure both unfolding speed and accuracy, this invention proposes a high-precision phase unfolding method based on fringe order correction, which can unfold high-frequency fringes with high accuracy without increasing the number of projection frames. Summary of the Invention

[0004] The purpose of this invention is to provide a high-precision phase unfolding method based on fringe order correction, which solves the problems of existing time phase unfolding methods and spatial phase unfolding methods. This invention improves the robustness of phase unfolding to defect areas through fringe order correction, unfolds high-frequency fringes with high precision without increasing the number of projection frames, and ensures unfolding speed and accuracy, providing strong support for achieving high-precision and high-speed 3D reconstruction.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A high-precision phase unfolding method based on fringe order correction is characterized by the following steps:

[0007] Step 1: Display dual-frequency stripes on the screen and illuminate the surface of the object to be tested, and simultaneously acquire images with a camera; the dual-frequency stripes are single-frequency stripes and high-frequency stripes, both of which are sinusoidal stripe structured light that satisfies 4-step phase shift;

[0008] Step 2: Calculate the principal phase value using the acquired fringe pattern to obtain the single-frequency wrapped phase φ1(x,y) and the high-frequency wrapped phase φ1(x,y). h (x,y);

[0009] Step 3: Calculate the fringe modulation index M(x,y) of the high-frequency stripes and use the modulation index image M(x,y) to complete the defect region identification and obtain the defect mask mask(x,y);

[0010] Step 4: Extend the single-frequency wrapped phase φ1(x,y) to a high-frequency period to obtain the reference phase φ. refer (x,y) and calculate the reference order k. refer (x,y);

[0011] Step 5: Unfold the high-frequency wrapped phase using the spatial phase unfolding method, and calculate the fringe order k based on the spatial phase unfolding results. s (x,y);

[0012] Step 6: In the defect area, according to the reference level k refer (x,y) represents the fringe order k s Sign correction is performed on (x,y) to obtain the corrected fringe order k. m (x,y);

[0013] Step 7: Calculate the reference phase φ in the defect area. refer (x,y) and the spatial phase expansion result Φ s The order difference k of (x,y) diff (x,y), and superimposed the corrected fringe order k m (x,y), to obtain the true fringe order k′ m (x,y); In the non-defect region, the stripe order k s (x,y) represents the true fringe order k′ m (x,y);

[0014] Step 8: Calculate k′ based on the actual stripe order. m (x,y) yields the high-precision phase unfolding result Φ based on fringe order correction. h (x,y).

[0015] Furthermore, in step 1, taking the fringes in the x-direction as an example, the single-frequency fringes and high-frequency fringes are specifically represented as follows:

[0016]

[0017] Where (x,y) represents pixel coordinates, I n (x,y) represents the intensity of the sinusoidal stripes displayed on the screen, A(x,y) represents the background light intensity, and B(x,y) reflects the contrast of the generated stripes; N=4, which represents four-step phase shift; f represents the stripe frequency. For single-frequency stripes: f=1, and for high-frequency stripes: f=F, where F represents the frequency of the high-frequency stripes.

[0018] Furthermore, in step 1, the image acquisition by the camera is specifically represented as follows:

[0019]

[0020] Among them, I n φ′(x,y) represents the light intensity distribution of the image captured by the camera, φ(x,y) represents the phase change introduced by the three-dimensional shape of the surface of the object under test, R(x,y) represents the surface reflectivity of the object under test, A′(x,y) is the measured background light intensity, and B′(x,y) reflects the contrast of the captured stripes.

[0021] Furthermore, in step 2, the calculation process for the wrapping phase is as follows:

[0022]

[0023] Among them, I n ′(x,y) represents the light intensity distribution in the image captured by the camera, n=1,2,3,4.

[0024] Furthermore, in step 3, the calculation process for the fringe modulation degree M(x,y) of the high-frequency fringes is as follows:

[0025]

[0026] Among them, I n ′(x,y) represents the light intensity distribution in the image captured by the camera, n=1,2,3,4;

[0027] By combining the modulation image to identify the defect region on the surface of the object under test, a defect mask mask(x,y) is obtained, where the defect region is represented as 1 and the non-defect region is represented as 0. Specifically, the defect identification method can use edge detection based on the Sobel operator, image segmentation algorithm, Fourier transform detection, or deep learning-based defect detection.

[0028] Furthermore, in step 4, the reference phase φ refer (x,y) specifically refers to:

[0029] φ refer (x,y)=F·φ1(x,y)

[0030] Where F represents the frequency of the high-frequency stripe;

[0031] Reference level k refer (x,y) specifically refers to:

[0032]

[0033] In this context, round[·] means rounding to the nearest integer.

[0034] Furthermore, in step 5, further, the stripe order k s (x,y) specifically refers to:

[0035]

[0036] Where, Φ s (x,y) represents the spatial phase expansion result.

[0037] Furthermore, in step 6, the corrected fringe order k m (x,y) specifically refers to:

[0038] k m (x,y)=abs(k s (x,y))·sign(k refer (x,y))

[0039] Where abs(·) represents taking the absolute value; sign(·) represents taking the sign.

[0040] Furthermore, in step 7, in the defect area, the level difference k diff (x,y) specifically refers to:

[0041]

[0042] True stripe order k′ m (x,y) specifically refers to:

[0043] k′ m (x,y)=k m (x,y)+k diff (x,y).

[0044] Furthermore, in step 8, the high-precision phase unfolding result Φ based on fringe order correction is... h (x,y) specifically refers to:

[0045] Φ h (x,y)=2πk′ m (x,y)+φ h (x,y)

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] This invention provides a high-precision phase unfolding method based on fringe order correction. The order correction process does not involve the direct use of single-frequency phase, but only uses it as a correction reference. This avoids the amplification of single-frequency phase noise in the traditional dual-frequency phase unfolding process, and can unfold higher frequency fringes without additionally projecting intermediate frequency fringes, thus achieving high-precision phase unfolding. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the high-precision phase unfolding method based on fringe order correction in this invention.

[0049] Figure 2 The result of the dual-frequency phase expansion in Comparative Example 1 is shown in the figure.

[0050] Figure 3 This is an error distribution diagram of the dual-frequency phase expansion in Comparative Example 1.

[0051] Figure 4 This is a diagram showing the results of phase expansion for the modulation system in Comparative Example 2.

[0052] Figure 5 This is an error distribution diagram of the phase expansion of the modulation order in Comparative Example 2.

[0053] Figure 6 This is a diagram showing the result of high-precision phase unfolding based on fringe order correction in an embodiment of the present invention.

[0054] Figure 7 This is an error distribution diagram of high-precision phase expansion based on fringe order correction in an embodiment of the present invention. Detailed Implementation

[0055] To make the invention's objectives, technical solutions, and effects clearer and more complete, the invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0056] This invention provides a high-precision phase unwrapping method based on fringe order correction. First, the image acquired by the camera is processed to obtain the principal phase values, namely the single-frequency wrapped phase φ1(x,y) and the high-frequency wrapped phase φ. h Simultaneously, the modulation index M(x,y) of the high-frequency fringes is calculated, and the defect region is identified using the modulation index image M(x,y), resulting in a defect mask mask(x,y). Then, the high-frequency wrapping phase is unfolded using a spatial phase unfolding method, and the result Φ is used to determine the defect region. s (x,y) Inverse calculation of stripe order k s (x,y); Extending the single-frequency wrapper phase to a high-frequency period, the reference order k is calculated. refer(x,y), and in the defect region according to the reference level k refer (x,y) represents the fringe order k s Sign correction is performed on (x,y) to obtain the corrected spatial phase expansion order k. m (x,y); Finally, calculate the order difference k between the reference phase and the spatial phase expansion result. diff (x,y), by the corrected spatial phase expansion order k m (x,y) Superposition order difference k diff (x,y) yields the true order of the defect region, and the phase expansion result Φ is calculated based on its true order. h (x,y); for non-defect regions, the order result of spatial phase expansion is used to ensure the smoothness of the expansion: directly using the fringe order k s The phase expansion result Φ is obtained by calculation (x,y). h (x,y). The order correction process of this invention does not involve the direct use of single-frequency phase, but only uses it as a correction reference, which avoids the amplification of single-frequency phase noise in the traditional dual-frequency phase unfolding process, and can unfold higher frequency fringes without additionally projecting intermediate frequency fringes, thus achieving high-precision phase unfolding.

[0057] Specifically, this embodiment provides a high-precision phase unfolding method based on fringe order correction, including the following steps:

[0058] Step 1: Display dual-frequency stripes on the surface of the object to be tested on the LCD screen, and simultaneously acquire images with the camera; the dual-frequency stripes are single-frequency stripes and high-frequency stripes, both of which are sinusoidal stripe structured light that satisfies 4-step phase shift.

[0059] The display screen shows single-frequency and high-frequency stripes in the x-direction according to the following rules:

[0060]

[0061] Where (x,y) represents pixel coordinates, I n (x,y) represents the intensity of the sinusoidal stripes displayed on the screen, A(x,y) represents the background light intensity, and B(x,y) reflects the contrast of the generated stripes; N=4, which represents four-step phase shift; f represents the stripe frequency. For single-frequency sinusoidal stripes: f=1; for high-frequency stripes: f=F, where F represents the frequency of the high-frequency stripes.

[0062]

[0063] Among them, I nφ′(x,y) represents the light intensity distribution of the image captured by the camera, φ(x,y) represents the phase change introduced by the three-dimensional shape of the surface of the object under test, R(x,y) represents the surface reflectivity of the object under test, A′(x,y) is the measured background light intensity, and B′(x,y) reflects the contrast of the captured stripes.

[0064] Step 2: Calculate the principal phase value using the acquired fringe image to obtain the single-frequency wrapped phase φ1(x,y) and the high-frequency wrapped phase φ1(x,y). h (x,y);

[0065] The specific calculation process for the wrapping phase is as follows:

[0066]

[0067] Step 3: Calculate the fringe modulation index M(x,y) of the high-frequency stripes and complete the defect region identification to obtain the defect mask mask(x,y);

[0068] The calculation process for the fringe modulation degree M(x,y) of high-frequency fringes is as follows:

[0069]

[0070] Further, by combining the modulation image M(x,y) to identify the defect region on the surface of the object under test, a defect mask mask(x,y) is obtained, where the defect region is represented as 1 and the non-defect region is represented as 0. Specifically, the defect identification method can use edge detection based on the Sobel operator, image segmentation algorithm, Fourier transform detection, or deep learning-based defect detection.

[0071] Step 4: Extend the single-frequency wrapper phase to a high-frequency period to obtain the reference phase φ. refer (x,y) and calculate the reference order k. refer (x,y);

[0072] Reference phase φ refer (x,y) specifically refers to:

[0073] φ refer (x,y)=F·φ1(x,y)

[0074] Where F represents the frequency of the high-frequency stripe;

[0075] Reference level k refer (x,y) specifically refers to:

[0076]

[0077] Where, round[·] means rounding to the nearest integer;

[0078] Step 5: Unfold the high-frequency wrapped phase using the spatial phase unfolding method, and calculate the fringe order k based on the spatial phase unfolding results. s (x,y);

[0079] The spatial phase unwrapping method is specifically a quality map-guided spatial phase unwrapping method based on modulation ordering. The spatial phase unwrapping result is Φ. s (x,y), then the fringe order k s (x,y) specifically refers to:

[0080]

[0081] Step 6: In the defect area, according to the reference level k refer (x,y) represents the fringe order k s Sign correction is performed on (x,y) to obtain the corrected fringe order k. m (x,y);

[0082] Corrected fringe order k m (x,y) specifically refers to:

[0083] k m (x,y)=abs(k s (x,y))·sign(k refer (x,y))

[0084] Where abs(·) represents taking the absolute value; sign(·) represents taking the sign;

[0085] Step 7: Calculate the reference phase φ in the defect area. refer (x,y) and the spatial phase expansion result Φ s The order difference k of (x,y) diff (x,y), and superimposed the corrected fringe order k m (x,y), to obtain the true fringe order k′ m (x,y); In the non-defect region, the fringe order k is calculated from the spatial phase unfolding result. s (x,y) represents the true fringe order k′ m (x,y);

[0086] In the defect region, the order difference k diff (x,y) specifically refers to:

[0087]

[0088] The true stripe order k′ in the defect region m (x,y) specifically refers to:

[0089] k′ m (x,y)=km (x,y)+k diff (x,y)

[0090] Step 8: Calculate k′ based on the actual stripe order. m (x,y) yields the high-precision phase unfolding result Φ based on fringe order correction. h (x,y);

[0091] High-precision phase unfolding results based on fringe order correction Φ h (x,y) specifically refers to:

[0092] Φ h (x,y)=2πk′ m (x,y)+φ h (x,y)

[0093] To illustrate the beneficial effects of the present invention, this embodiment also compares the results of the traditional dual-frequency phase expansion method and the quality map-guided phase expansion method based on modulation ordering. Example 1: The dual-frequency phase expansion method is used as a comparison, such as... Figure 2 The figure shown is the result of the dual-frequency phase expansion in Example 1, as follows: Figure 3 The figure shown is the error distribution diagram of the dual-frequency phase expansion in Comparative Example 1, with a root mean square error of 2.34 rad; Figure 4 The figure shown is the result of phase expansion of the modulation order in Comparative Example 2, as follows: Figure 5 The figure shown is the error distribution diagram of the modulation ordering phase expansion in Comparative Example 2, with a root mean square error of 0.28 rad. Figure 6 The image shown is a high-precision phase unfolding result based on fringe order correction in an embodiment of the present invention, as follows: Figure 7 The figure shows the error distribution of high-precision phase unfolding based on stripe order correction in an embodiment of the present invention, with a root mean square error of 0.13 rad. As can be seen from the figure, the high-precision phase unfolding method proposed in this invention significantly reduces the error in the phase unfolding process, especially recovers the phase of the defect region better, and does not require additional stripe projection, thus improving the measurement speed.

[0094] The above description is merely a specific embodiment of the present invention. Any feature disclosed in this specification may be replaced by other equivalent or similar features unless otherwise specified. All disclosed features, or steps in all methods or processes, may be combined in any way except for mutually exclusive features and / or steps.

Claims

1. A high-precision phase unwrapping method based on fringe order correction, characterized in that, The method comprises the following steps: Step 1, display double-frequency fringe irradiation to the surface of the object to be measured through a display screen, and synchronously collect images by a camera; the double-frequency fringe is single-frequency fringe with a frequency of 1 and high-frequency fringe with a frequency of F, both of which are sinusoidal fringe structured light satisfying 4-step phase shift; Step 2, the phase principal value is calculated by using the collected fringe image, and single-frequency fringe wrapped phase Φ1(x, y) and high-frequency fringe wrapped phase Φ2(x, y) are obtained respectively. h (x,y) Step 3, calculate fringe modulation M(x, y) of the high-frequency fringe and complete defect area identification by using the modulation image to obtain a defect mask mask(x, y); Step 4, extend the single frequency wrapped phase Φ1(x, y) to high frequency period to get the reference phase Φ refer (x, y), and calculate the reference order k refer (x, y); Step 5, combine the fringe modulation M(x, y), expand the high frequency wrapped phase using spatial phase unwrapping method, and based on the spatial phase unwrapping result Φ s (x, y) to calculate the fringe order k s (x, y); Step 6, sign correction of (x, y) pair of stripe level k refer (x,y) pair of stripe level k s (x,y) pair of stripe level k m (x,y) pair of stripe level k Step 7, in defect area, calculate reference phase Φ refer (x,y) and spatial phase expansion result Φ s (x,y) and superimpose corrected fringe order k diff (x,y) and superimpose corrected fringe order k m (x,y) to get real fringe order k' m (x,y); in non-defect area, take fringe order k s (x,y) as real fringe order k' m (x,y) Step 8, Calculate k' from the true fringe order m (x,y) to get the high precision phase unwrapping result Φ based on fringe order correction h (x,y).

2. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 1, taking the fringe in the x direction as an example, the single-frequency fringe and the high-frequency fringe are specifically represented as: where (x, y) represents pixel coordinates, I n (x,y) represents the intensity of the displayed sinusoidal fringe, A(x,y) represents the background light intensity, B(x,y) reflects the contrast of the generated fringe; N=4 represents four-step phase shift; f represents the fringe frequency, for single frequency fringe: f=1, for high frequency fringe: f=F, F represents the frequency of the high frequency fringe.

3. The high-precision phase unwrapping method based on fringe order correction according to claim 2, characterized in that, In step 1, the image collected by the camera is specifically represented as: where I n (x, y) represents the light intensity distribution of the image captured by the camera, Φ(x, y) represents the phase change amount introduced by the three-dimensional topography of the surface of the object to be measured; R(x, y) represents the reflectivity of the surface of the object to be measured, A'(x, y) is the measured background light intensity, and B'(x, y) reflects the contrast of the captured fringe.

4. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 2, the calculation process of the wrapped phase is specifically as follows: where I n (x, y) represents the light intensity distribution of the image captured by the camera, n = 1, 2, 3, 4.

5. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 3, the calculation process of the fringe modulation M(x, y) of the high-frequency fringe is as follows: where I n (x, y) represents the light intensity distribution of the image captured by the camera, n = 1, 2, 3, 4; In combination with the modulation image M(x, y), the defect area on the surface of the object to be measured is identified to obtain a defect mask mask(x, y), wherein the defect area is represented as 1 and the non-defect area is represented as 0.

6. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 4, the reference phase Φ refer (x,y) is given by: Φ refer (x,y) = F - Φ1(x,y) Wherein, F represents the frequency of the high-frequency fringe; Reference order k refer (x, y) is specified as: Wherein, round[·] represents a rounding function.

7. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 5, the stripe order k s (x,y) is given by: where Φ s (x,y) represents the spatial phase unwrapping result.

8. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 6, the corrected fringe order k m (x,y) is given by: k m (x,y) = abs(k s (x,y)) · sign(k refer (x,y)) Wherein, abs(·) represents an absolute value function; sign(·) represents a sign function.

9. The high-precision phase unwrapping method based on fringe order correction according to claim 1, wherein, In step 7, in the defective area, the level difference k diff (x, y) is specifically: Real stripe order k' m (x, y) is given by: k' m (x,y) = k m (x,y) + k diff (x,y).

10. The method of claim 1, wherein the method is a high-precision phase unwrapping method based on fringe order correction. In step 8, the high-precision phase unwrapping result Φ based on fringe order correction h (x, y) is specifically: Φ h (x,y) = 2πk' m (x,y) + φ h (x,y).

Images