High-precision phase unwrapping method based on stripe level correction

Abstract

The invention belongs to the field of visual inspection, and provides a high-precision phase unwrapping method based on stripe level correction, which is used for solving the problems of the existing time phase unwrapping method and space phase unwrapping method. The method comprises the following steps: firstly, processing an image acquired by a camera to obtain single-frequency and high-frequency wrapped phases, and simultaneously completing defect area identification; a high-frequency wrapped phase is unwrapped by adopting a space phase unwrapping method, and the fringe order is inversely calculated; carrying out single-frequency wrapped phase expansion to calculate a reference level, and carrying out symbol correction on a stripe level in a defect area; and finally, the order difference of the unwrapping result of the reference phase and the space phase is calculated, the stripe order after superposition correction is used as the real order of the defect area, the stripe order is used as the real order of the non-defect area, and the phase unwrapping result is calculated according to the real order. According to the method, the robustness of phase unwrapping to a defect area is improved through stripe level correction, and high-frequency stripes are unwrapped with high precision under the condition that the number of projection frames is not increased.

Description

Technical Field

[0001] The invention belongs to the field of visual detection, and specifically provides a high-precision phase unwrapping method based on fringe order correction. Background Art

[0002] In the field of three-dimensional measurement based on phase information, phase unwrapping technology, as a core link, has long been widely concerned by academia and industry. In the process of phase unwrapping, the continuous absolute phase value is restored by analyzing the phase jumps in the wrapped phase image, providing accurate phase data support for subsequent three-dimensional reconstruction; the accuracy of phase unwrapping directly determines the final accuracy of three-dimensional measurement, and its computational efficiency directly affects the real-time performance of the entire three-dimensional reconstruction system. In application scenarios such as industrial inspection, biomedical imaging, and digital protection of cultural heritage, higher and higher requirements are placed on the accuracy and speed of three-dimensional reconstruction; therefore, the development of efficient and robust phase unwrapping methods to achieve high-precision and high-speed three-dimensional reconstruction has become an important research topic in the field of three-dimensional measurement.

[0003] As the two most commonly used phase unwrapping methods, temporal phase unwrapping and spatial phase unwrapping have their own advantages and disadvantages. Temporal phase unwrapping has the advantages of high unwrapping accuracy and no error propagation along the unwrapping path, but it often requires the projection of multiple sets of fringes with different frequencies, and the multi-frequency phase unwrapping method has the risk of amplifying errors. Spatial phase unwrapping only requires fringes of a single frequency, which is suitable for unwrapping smooth areas, but is limited by the assumption of surface smoothness, and errors will propagate along the unwrapping path. In order to overcome the shortcomings of the two methods and strive to ensure both unwrapping speed and accuracy, the present invention proposes a high-precision phase unwrapping method based on fringe order correction, which can unwrap high-frequency fringes with high accuracy without increasing the number of projection frames. Summary of the invention

[0004] The purpose of the present invention is to provide a high-precision phase unwrapping method based on fringe order correction, so as to solve the problems existing in the existing time phase unwrapping method and the spatial phase unwrapping method. The present invention improves the robustness of phase unwrapping to defective areas through the correction of fringe order, and unwraps high-frequency fringes with high precision without increasing the number of projection frames, while ensuring the unwrapping speed and accuracy, providing strong support for achieving high-precision and high-speed three-dimensional reconstruction.

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

[0006] A high-precision phase unwrapping method based on fringe order correction, characterized in that it comprises the following steps:

[0007] Step 1, displaying dual-frequency stripes on the display screen and irradiating the surface of the object to be measured, and synchronously collecting images by the camera; the dual-frequency stripes are single-frequency stripes and high-frequency stripes, both of which are sinusoidal stripe structured light that satisfies a 4-step phase shift;

[0008] Step 2: Use the collected fringe pattern to calculate the main phase value and obtain the single-frequency wrapped phase φ 1 (x,y) and high frequency wrapping phase φ h (x,y);

[0009] Step 3, calculate the fringe modulation degree M(x, y) of the high-frequency fringe and use the modulation degree image M(x, y) to complete the defect area recognition and obtain the defect mask mask(x, y);

[0010] Step 4: Wrap the single frequency phase φ 1 (x,y) is expanded to a high frequency period, and the reference phase φ is obtained refer (x,y), and calculate the reference level k refer (x,y);

[0011] Step 5: Use the spatial phase unwrapping method to unwrap the high-frequency wrapped phase, and inversely calculate the fringe order k based on the spatial phase unwrapping result. s (x,y);

[0012] Step 6: In the defect area, according to the reference level k refer (x,y) for fringe level k s (x, y) is sign corrected to obtain the corrected fringe level k m (x,y);

[0013] Step 7: Calculate the reference phase φ in the defect area refer (x,y) and spatial phase unwrapping result Φ s The order difference k of (x,y) diff (x, y), and superimpose the corrected fringe level k m (x,y), get the real stripe level k′ m (x, y); in the non-defective area, the fringe level k s (x,y) is the real stripe level k′ m (x,y);

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

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

[0016]

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

[0018] Furthermore, in step 1, the image captured by the camera is specifically expressed as:

[0019]

[0020] Among them, I n ′(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 morphology 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 collected fringes.

[0021] Furthermore, in step 2, the calculation process of the wrapped phase is specifically as follows:

[0022]

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

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

[0025]

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

[0027] The defect area on the surface of the object to be tested is identified by combining the modulation image to obtain a defect mask (x, y), where the defect area is represented by 1 and the non-defect area is represented by 0. Specifically, the defect identification method can use edge detection based on the Sobel operator, image segmentation algorithm, Fourier transform detection or defect detection based on deep learning.

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

[0029] φ refer (x,y)=F·φ1 (x,y)

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

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

[0032]

[0033] Among them, round[·] means rounding to the nearest integer.

[0034] Further, in step 5, further, the stripe level k s (x,y) is specifically:

[0035]

[0036] Among them, Φ s (x,y) represents the spatial phase unwrapping result.

[0037] Furthermore, in step 6, the corrected fringe level k m (x,y) is specifically:

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

[0039] Among them, abs(·) means taking the absolute value; sign(·) means taking the sign.

[0040] Further, in step 7, in the defect area, the order difference k diff (x,y) is specifically:

[0041]

[0042] True fringe level k′ m (x,y) is specifically:

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

[0044] Further, in step 8, the high-precision phase unwrapping result Φ based on fringe order correction is h (x,y) is specifically:

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

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention provides a high-precision phase unwrapping method based on fringe order correction. The order correction process does not involve direct use of single-frequency phase, but only serves as a correction reference, thereby avoiding the amplification of single-frequency phase noise in the traditional dual-frequency phase unwrapping process. In addition, higher-frequency fringes can be unwrapped without additionally projecting intermediate-frequency fringes, thereby achieving high-precision phase unwrapping. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic flow chart of the high-precision phase unwrapping method based on fringe order correction in the present invention.

[0050] Figure 2 This is the result diagram of dual-frequency phase unwrapping in comparative example 1.

[0051] Figure 3 This is the error distribution diagram of the dual-frequency phase unwrapping in comparative example 1.

[0052] Figure 4 This is the result diagram of the modulation degree sorting phase unwrapping in comparative example 2.

[0053] Figure 5 This is the error distribution diagram of the modulation degree sorting phase unwrapping in comparative example 2.

[0054] Figure 6 This is a result diagram of high-precision phase unwrapping based on fringe order correction in an embodiment of the present invention.

[0055] Figure 7 This is an error distribution diagram of high-precision phase unwrapping based on fringe order correction in an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical scheme and technical effect of the present invention more clear and complete, the present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0057] The present invention provides a high-precision phase unwrapping method based on fringe order correction. First, the image collected by the camera is processed to obtain the main phase value, which is the single-frequency wrapped phase φ 1 (x,y) and high frequency wrapping phase φ h (x, y), and at the same time, calculate the fringe modulation M(x, y) of the high-frequency fringe and use the modulation image M(x, y) to complete the defect area recognition and obtain the defect mask mask(x, y); then, the spatial phase unwrapping method is used to unwrap the high-frequency wrapped phase, and based on the result of the spatial phase unwrapping Φ s (x,y) back-calculate fringe level ks (x,y); expand the single-frequency wrapped phase to the high-frequency period and calculate the reference order k refer (x, y), and in the defect area according to the reference level k refer (x,y) for fringe level k s (x, y) is sign corrected to obtain the corrected spatial phase unwrapping order k m (x, y); Finally, calculate the order difference k between the reference phase and the spatial phase unwrapping result diff (x, y), which is the corrected spatial phase unwrapping level k m (x,y) superposition order difference k diff (x, y) to obtain the true order of the defect area, and calculate the phase unwrapping result Φ based on its true order h (x, y); for non-defective areas, the order result of spatial phase unwrapping is used to ensure the smoothness of the unwrapping: directly using the fringe order k s (x,y) calculates the phase unwrapping result Φ h (x, y). The order correction process of the present invention does not involve direct use of the single-frequency phase, but only serves as a correction reference, thereby avoiding the amplification of the single-frequency phase noise in the traditional dual-frequency phase unwrapping process, and can unwrap higher-frequency fringes without additionally projecting intermediate-frequency fringes, thereby achieving high-precision phase unwrapping.

[0058] Specifically, this embodiment provides a high-precision phase unwrapping method based on fringe order correction, comprising the following steps:

[0059] Step 1, displaying dual-frequency stripes on an LCD screen to illuminate the surface of the object to be measured, and synchronously collecting 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 a 4-step phase shift;

[0060] The display screen displays single-frequency stripes and high-frequency stripes in the x-direction according to the following rules:

[0061]

[0062] Where (x, y) represents the pixel coordinates, I n (x, y) represents the intensity of the sinusoidal stripes displayed on the display screen, A(x, y) represents the background light intensity, and B(x, y) reflects the contrast of the generated stripes; N = 4, indicating a 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.

[0063]

[0064] Among them, I n′(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 morphology 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 collected fringes.

[0065] Step 2: Use the collected fringe image to calculate the phase principal value and obtain the single-frequency wrapped phase φ 1 (x,y) and high frequency wrapping phase φ h (x,y);

[0066] The calculation process of the wrapped phase is as follows:

[0067]

[0068] Step 3, calculate the fringe modulation degree M(x,y) of the high-frequency fringe and complete the defect area identification to obtain the defect mask mask(x,y);

[0069] The calculation process of the fringe modulation degree M(x,y) of the high-frequency fringe is:

[0070]

[0071] The modulation image M(x, y) is further combined to identify the defect area on the surface of the object to be tested, and a defect mask mask(x, y) is obtained, where the defect area is represented by 1 and the non-defect area is represented by 0. Specifically, the defect recognition method can use edge detection based on the Sobel operator, image segmentation algorithm, Fourier transform detection or defect detection based on deep learning.

[0072] Step 4: Expand the single-frequency wrapped phase to the high-frequency period to obtain the reference phase φ refer (x,y), and calculate the reference level k refer (x,y);

[0073] Reference phase φ refer (x,y) is specifically:

[0074] φ refer (x,y)=F·φ 1 (x,y)

[0075] Where F represents the frequency of high-frequency fringes;

[0076] Reference level k refer (x,y) is specifically:

[0077]

[0078] Among them, round[·] means rounding to the nearest integer;

[0079] Step 5: Use the spatial phase unwrapping method to unwrap the high-frequency wrapped phase, and inversely calculate the fringe order k based on the spatial phase unwrapping result. s (x,y);

[0080] 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 level k s (x,y) is specifically:

[0081]

[0082] Step 6: In the defect area, according to the reference level k refer (x,y) for fringe level k s (x, y) is sign corrected to obtain the corrected fringe level k m (x,y);

[0083] Corrected fringe level k m (x,y) is specifically:

[0084] k m (x,y)=abs(k s (x,y))·sign(k refer (x,y)

[0085] Among them, abs(·) means taking the absolute value; sign(·) means taking the sign;

[0086] Step 7: Calculate the reference phase φ in the defect area refer (x,y) and spatial phase unwrapping result Φ s The order difference k of (x,y) diff (x, y), and superimpose the corrected fringe level k m (x,y), get the real stripe level k′ m (x, y); In the non-defective area, the fringe order k calculated by the spatial phase unwrapping result s (x,y) is the real stripe level k′ m (x,y);

[0087] In the defect area, the order difference k diff (x,y) is specifically:

[0088]

[0089] The actual fringe order k′ of the defect area m (x,y) is specifically:

[0090] k′ m(x,y)=k m (x,y)+k diff (x,y)

[0091] Step 8: Calculate k′ based on the actual fringe level m (x,y) to obtain the high-precision phase unwrapping result Φ based on fringe order correction h (x,y);

[0092] High-precision phase unwrapping results based on fringe order correction Φ h (x,y) is specifically:

[0093] Φ h (x,y)=2πk′ m (x,y)+φ h (x,y)

[0094] In order to illustrate the beneficial effects of the present invention, this embodiment also uses the results of the traditional dual-frequency phase unwrapping method and the quality map guided phase unwrapping method based on modulation ordering as comparison. Example 1: The dual-frequency phase unwrapping method is used as a comparison. Figure 2 The result of dual-frequency phase unwrapping in Example 1 is shown in FIG. Figure 3 The error distribution diagram of the dual-frequency phase unwrapping in comparative example 1 is shown, and the root mean square error is 2.34 rad; Figure 4 The result of the modulation order phase unwrapping in comparative example 2 is shown as follows: Figure 5 The error distribution diagram of the modulation order phase unwrapping in comparative example 2 is shown, and the root mean square error is 0.28 rad. Figure 6 FIG. 4 is a diagram showing the result of high-precision phase unwrapping based on fringe order correction in an embodiment of the present invention. Figure 7 The figure shows the error distribution diagram of the high-precision phase unwrapping based on fringe order correction in an embodiment of the present invention, and the root mean square error is 0.13 rad; it can be seen from the figure that the high-precision phase unwrapping method proposed in the present invention significantly reduces the error in the phase unwrapping process, especially better restores the phase of the defective area, and does not require additional projected fringes, thereby improving the measurement speed.

[0095] The above description is only a specific implementation mode of the present invention. Any feature disclosed in this specification, unless otherwise stated, can be replaced by other alternative features that are equivalent or have similar purposes; all the disclosed features, or all the steps in the methods or processes, except for mutually exclusive features and / or steps, can be combined in any way.

Claims

1. A high-precision phase unwrapping method based on fringe order correction, characterized in that: The following steps are involved: Step 1, displaying dual-frequency stripes on the display screen and irradiating the surface of the object to be measured, and synchronously collecting images by the camera; the dual-frequency stripes are single-frequency stripes with a frequency of 1 and high-frequency stripes with a frequency of F, both of which are sinusoidal stripe structured light that satisfies a 4-step phase shift; Step 2: Use the collected fringe image to calculate the phase principal value, and obtain the single-frequency fringe package phase φ1(x, y) and the high-frequency fringe package phase φ h (x,y); Step 3, calculate the fringe modulation M(x,y) of the high-frequency fringe and use the modulation image to complete the defect area recognition, and obtain the defect mask mask(x,y); Step 4: Expand the single-frequency wrapped phase φ1(x,y) to the high-frequency period to obtain the reference phase φ refer (x,y), and calculate the reference level k refer (x,y); Step 5: Combined with the fringe modulation degree M(x,y), the spatial phase unwrapping method is used to unwrap the high-frequency wrapped phase, and based on the spatial phase unwrapping result Φ s (x,y) back-calculate fringe level k s (x,y); Step 6: In the defect area, according to the reference level k refer (x,y) for fringe level k s (x, y) is sign corrected to obtain the corrected fringe level k m (x,y); Step 7: Calculate the reference phase φ in the defect area refer (x,y) and spatial phase unwrapping result Φ s The order difference k of (x,y) diff (x, y), and superimpose the corrected fringe level k m (x,y), get the real stripe level k′ m (x, y); in the non-defective area, the fringe level k s (x,y) is the real stripe level k′ m (x,y); Step 8: Calculate k′ based on the actual fringe level m (x,y) to obtain 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, characterized in that: In step 1, taking the stripes in the x direction as an example, the single-frequency stripes and the high-frequency stripes are specifically expressed as: Where (x, y) represents the pixel coordinates, I n (x, y) represents the intensity of the sinusoidal stripes displayed on the display screen, A(x, y) represents the background light intensity, and B(x, y) reflects the contrast of the generated stripes; N = 4, indicating a four-step phase shift; f represents the stripe frequency, for single-frequency stripes: f = 1, for high-frequency stripes: f = F, and F represents the frequency of the high-frequency stripes.

3. The high-precision phase unwrapping method based on fringe order correction according to claim 2, characterized in that: In step 1, the camera captures the image as follows: Among them, I n ′(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 morphology 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 collected fringes.

4. The high-precision phase unwrapping method based on fringe order correction according to claim 1, characterized in that: In step 2, the calculation process of the wrapped phase is as follows: Among them, 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, characterized in that: In step 3, the calculation process of the fringe modulation degree M(x,y) of the high-frequency fringe is: Among them, I n ′(x,y) represents the light intensity distribution of the image collected by the camera, n=1,2,3,4; Combined with the modulation image M(x, y), the defect area on the surface of the object to be tested is identified to obtain the defect mask mask(x, y), where the defect area is represented by 1 and the non-defect area is represented by 0.

6. The high-precision phase unwrapping method based on fringe order correction according to claim 1, characterized in that: In step 4, the reference phase φ refer (x,y) is specifically: φ refer (x,y)=F·φ1(x,y) Where F represents the frequency of high-frequency fringes; Reference level k refer (x,y) is specifically: Where round[·] represents the rounding function.

7. The high-precision phase unwrapping method based on fringe order correction according to claim 1, characterized in that: In step 5, the stripe level k s (x,y) is specifically: Among them, 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, characterized in that: In step 6, the corrected fringe level k m (x,y) is specifically: k m (x,y)=abs(k s (x,y))·sign(k refer (x,y)) Among them, abs(·) represents the absolute value function; sign(·) represents the sign function.

9. The high-precision phase unwrapping method based on fringe order correction according to claim 1, characterized in that: In step 7, in the defect area, the order difference k diff (x,y) is specifically: True fringe level k′ m (x,y) is specifically: k′ m (x,y)=k m (x,y)+k diff (x,y)。 10. The high-precision phase unwrapping method based on fringe order correction according to claim 1, characterized in that: In step 8, the high-precision phase unwrapping result Φ based on fringe order correction is h (x,y) is specifically: Φ h (x,y)=2πk′ m (x,y)+φ h (x,y)。

Images