A high dynamic range fringe projection 3D measurement method based on fringe restoration

Abstract

The present invention provides a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration, comprising the following steps: projecting a multi-step grayscale sinusoidal fringe pattern onto a highly reflective surface; dividing the captured image into a reliable area, a light saturation area, and a deep saturation area; calculating the A and B surfaces of the reliable area using the Euler formula method; finding the unsaturated lower intersection point between adjacent fringes to expand F lower The CSI-BSI interpolation method is used to complete the shallow and deep saturation areas of the three surfaces, and based on the relationship between them, the accurate A is calculated for each position. The saturated fringes are then repaired and the phase is further calculated. This invention provides a high-dynamic range fringe projection 3D measurement method based on fringe repair. This method can effectively repair saturated fringes and improve the dynamic range of the system. Using the fringe repair method, the information utilization rate of each image is increased, the number of projection patterns required for high-dynamic range measurement is reduced, and the measurement speed is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of structured light three-dimensional measurement, and in particular to a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration. Background Art

[0002] With the development of science and technology, the three-dimensional information of objects has received more and more attention. Fringe projection three-dimensional measurement technology is widely used in industrial modeling, cultural relics protection, biomedicine, intelligent driving and other fields.

[0003] As structured light 3D measurement is increasingly used, the characteristics of the objects being measured are becoming more complex. The objects that need to be measured are no longer limited to white or light-colored objects. Sometimes, it is also necessary to measure objects with large changes in surface reflectivity.

[0004] Due to the limited dynamic range of the camera, the captured fringe pattern will be saturated. If the phase is calculated directly after the fringe is saturated, it will deviate from the ideal phase, resulting in inaccurate three-dimensional measurement.

[0005] Therefore, it is necessary to provide a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration to solve the above technical problems. Summary of the Invention

[0006] The present invention provides a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration, which solves the problem that the dynamic range of current cameras is limited, the captured fringe pattern will be saturated, and if the phase is directly calculated after the fringe is saturated, it will deviate from the ideal phase, resulting in inaccurate three-dimensional measurement.

[0007] To solve the above technical problems, the present invention provides a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration, comprising the following steps:

[0008] S1, projecting a multi-step grayscale sinusoidal fringe pattern onto a highly reflective surface, and the CCD captures the multi-step phase-shifted, deformed, saturated fringes;

[0009] S2. Divide the captured image into reliable areas, shallow saturation areas, and deep saturation areas. Calculate the reliable areas using a multi-step phase-shift saturation fringe pattern. Calculate the A and B parameters of the reliable areas using the Euler formula method.

[0010] S3. Since the shallow saturation area and the deep saturation area cannot be perfectly solved, all reliable areas are expanded and eroded to obtain the CSI and BSI areas of A and B to roughly correspond to the shallow saturation area and the deep saturation area of ​​A and B. The CSI and BSI interpolation methods are used to interpolate and complete the CSI and BSI areas of A and B.

[0011] S4. Find the unsaturated lower intersection points between adjacent stripes. The surface where these points are located is Flower , using these lower intersection points as F lower reliable points, and use A, B and F lower The relationship between the two points is expanded by using A and B of the reliable point position. lower The reliable point, finally use F lower The extended reliable region is expanded and eroded to obtain F lower The CSI and BSI areas roughly correspond to F lower The shallow saturation area and deep saturation area are obtained by interpolating the CSI and BSI regions. lower Perform interpolation and completion;

[0012] S5. According to A, B and F lower The relationship between the three surfaces, A, B and F obtained by interpolation lower Calculate the exact A for each position;

[0013] S6. Using the relationship between stripes, the saturated stripes are repaired using the repaired A and the unsaturated stripes. After the saturated portion at the top of the stripe is repaired, the remaining portion is completed using CSI interpolation.

[0014] S7, when all fringes are repaired, use the multi-step phase shift method to calculate the phase;

[0015] S8. Use the complementary Gray code method to solve the order and unfold the wrapped phase.

[0016] Preferably, the S1 includes the following specific steps:

[0017] S1, projecting a grayscale sinusoidal fringe pattern onto a highly reflective surface, the CCD collects the deformed saturated fringes, the projected N fringe patterns have a constant phase difference δ, and the captured image also has a phase difference of δ, that is,

[0018]

[0019] in, N is the number of phase shift steps, which is an integer greater than or equal to 3, n=1, 2, ..., N is the phase shift sequence, A(x, y) is the average intensity, and B(x, y) is the intensity modulation.

[0020] Preferably, S2 includes the following specific steps:

[0021] S21. For each point in the image, divide it into a reliable area, a shallow saturation area, and a deep saturation area according to the saturation of the N phase shift images. If the values ​​of three or more of the N phase shift images at a certain point are less than 255, then this point is a reliable area, and the rest are unreliable areas. In the unreliable area, assume that A a is the actual A parameter, then when Aa When A<255, the point is a shallow saturation area. a When ≥255, this point is in the deep saturation area.

[0022] S22. The A parameter of the reliable area can be solved directly. The A parameter of the shallow saturation area and the deep saturation area cannot be solved directly. The formula for solving the A parameter of the reliable area is as follows:

[0023] According to Euler's formula e ix =cosx+isinx, I n (x,y) can be written as:

[0024]

[0025] The stripe subscripts are n1,n2,...,n k The stripes are not saturated. Combine the equations for these unsaturated stripes at the (x,y) position:

[0026]

[0027] Then the formula for solving the A and B parameters is:

[0028]

[0029] Preferably, S3 includes the following specific steps:

[0030] S31, after solving the reliable regions A and B, the reliable regions are expanded. The expansion distance should be more than half of the fringe period to obtain the expanded region;

[0031] Then the expanded area is eroded, and the erosion distance should be greater than the expansion distance to obtain the CSI area;

[0032] Finally, the CSI region is inverted to obtain the BSI region. Since the shallow saturation region and the deep saturation region cannot be perfectly solved, the CSI and BSI regions are obtained to roughly correspond to the shallow saturation region and the deep saturation region;

[0033] S32. Using the A values ​​of all reliable points in the CSI area as a basis, interpolate and complete the A value of the entire CSI area.

[0034] Interpolate and complete the B value of the entire CSI area based on the B values ​​of all reliable points in the CSI area;

[0035] The interpolation method uses the cubic spline interpolation method: cubic spline interpolation, referred to as CSI interpolation method;

[0036] S33, using the A values ​​of all reliable points in the BSI area as a basis, interpolating and completing the A value of the entire BSI area;

[0037] The B values ​​of all reliable points in the BSI area are used as the basis to interpolate and complete the B values ​​of the entire BSI area;

[0038] The interpolation method uses the biharmonic spline interpolation method: biharmonic spline interpolation, referred to as BSI interpolation method.

[0039] Preferably, the S4 includes the following specific steps:

[0040] S41、F lower The method for finding reliable points is as follows: the reliable point is the lower intersection point of adjacent stripes without saturation and the lower reliable point calculated by the reliable area of ​​A and B. n (x,y) and I n+1 (x,y), the equation of the intersection point is:

[0041]

[0042] Solving the above equation, the phase of the intersection position is:

[0043]

[0044] The fringe intensity value at the intersection position is:

[0045]

[0046] k is an integer. In order to determine whether these intersection points are lower intersection points, use I n (x,y) and I n+1 The intensity of the intersection of the (x,y) symmetrical stripes at that position is used to assist in the determination;

[0047] I n (x,y) and I n+1 The symmetrical stripes of (x,y) are represented as, MOD(N / 2+n) and I MOD(N / 2+n+1) , where 'MOD' borrows the concept of the remainder function and is expressed as:

[0048]

[0049] N represents the total number of phase shift steps, and the intensity of the symmetrical fringe intersection is expressed as:

[0050]

[0051] determination The point is the lower intersection point P lowerint The conditions are:

[0052]

[0053]

[0054] All F lower All lower intersection points of are the union of all adjacent stripe intersection points:

[0055]

[0056] In addition, A and B at the reliable position can also be used to control F lower The reliable points are further expanded, and the expanded formula is:

[0057]

[0058] S42, solve to F lower After the reliable area is obtained, the reliable area is expanded, and the expansion distance should be more than half of the stripe period to obtain the expanded area;

[0059] Then the expanded area is eroded, and the erosion distance should be greater than the expansion distance to obtain the CSI area;

[0060] Finally, the CSI region is inverted to obtain the BSI region;

[0061] Since the shallow saturation region and the deep saturation region cannot be solved perfectly, the CSI and BSI regions are obtained to roughly correspond to the shallow saturation region and the deep saturation region;

[0062] S43, using F of all reliable points in the CSI area lower The value is based on the F of the entire CSI area lower The value is interpolated and completed;

[0063] The interpolation method uses the cubic spline interpolation method: cubic spline interpolation, referred to as CSI interpolation method;

[0064] S44, use F of all reliable points in the BSI area lower The value is based on the F of the entire BSI area lower The interpolation method uses the biharmonic spline interpolation method: biharmonic spline interpolation, referred to as BSI interpolation method.

[0065] Preferably, the S5 includes the following specific steps:

[0066] S51, interpolated surfaces A, B and F lowerIn the figure, the strength of A and B is higher, the interpolation base points are few, and the error with the ideal surface is large. lower The interpolation base points are many and uniform, and the error with the ideal surface is small. The interpolated surfaces A, B, and F lower Written as A′, B′, F lo ' wer , the ideal surfaces are A, B, F lower , because the interpolation function can preserve the gradient when interpolating two sets of points with different intensities of the same trend, the ratio of A′, B′ is very close to the ratio of A, B:

[0067]

[0068] Therefore, the surfaces A″ and B″ with very small errors from the ideal surfaces A and B can be solved:

[0069]

[0070] Preferably, the S6 includes the following specific steps:

[0071] S61, when an image is saturated in N-step phase shift, the fringe repair value I of the saturated area n r The repair formula is as follows:

[0072] When N is an even number:

[0073]

[0074] When N is an odd number:

[0075]

[0076] S62. After the saturated phase-shifted image is repaired using the above repair formula, if there are still saturated positions, these positions are interpolated and completed using the CSI interpolation method based on the global unsaturated fringe values ​​and the repaired fringe values.

[0077] Preferably, the S7 includes the following specific steps:

[0078] S71. The wrapped phase can be solved by repairing N fringe patterns:

[0079]

[0080] The phase obtained by the above formula is the wrapped phase. To obtain height information, the unwrapped phase needs to be used. The absolute phase method can unwrap the wrapped phase.

[0081] Preferably, the S8 includes the following specific steps:

[0082] S81, projecting multiple Gray code fringe patterns onto the surface of a high reflectivity object, and using a CCD to obtain Gray code fringe patterns. The first few patterns are used to calculate the order, and the last pattern is used to correct the order to obtain the order K of the fringe.

[0083] S82. Use the order K to unfold the wrapped phase. The unfolding formula is as follows:

[0084]

[0085] The final Φ(x,y) obtained is the unfolded phase.

[0086] Compared with related technologies, the high dynamic range fringe projection 3D measurement method based on fringe restoration provided by the present invention has the following beneficial effects:

[0087] The present invention provides a high dynamic range fringe projection three-dimensional measurement method based on fringe repair. The method can effectively repair saturated fringes and improve the dynamic range of the system. Using the fringe repair method, the information utilization rate of each image is increased, the number of projection patterns required for high dynamic range measurement is reduced, and the measurement speed is effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 A schematic structural diagram of a preferred embodiment of a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration provided by the present invention;

[0089] Figure 2 Schematic diagram for calculating the intersection points of adjacent stripes;

[0090] Figure 3 Comparison diagram of the phase calculated before and after fringe repair. DETAILED DESCRIPTION

[0091] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0092] First embodiment

[0093] Please refer to Figure 1 、 Figure 2 and Figure 3 ,in, Figure 1 A schematic structural diagram of a preferred embodiment of a high dynamic range fringe projection three-dimensional measurement method based on fringe restoration provided by the present invention; Figure 2 Schematic diagram for calculating the intersection points of adjacent stripes; Figure 3 A high dynamic range fringe projection 3D measurement method based on fringe restoration includes the following steps:

[0094] S1, projecting a multi-step grayscale sinusoidal fringe pattern onto a highly reflective surface, and the CCD captures the multi-step phase-shifted, deformed, saturated fringes;

[0095] S2. Divide the captured image into reliable areas, shallow saturation areas, and deep saturation areas. Calculate the reliable areas using a multi-step phase-shift saturation fringe pattern. Calculate the A and B parameters of the reliable areas using the Euler formula method.

[0096] S3. Since the shallow saturation area and the deep saturation area cannot be perfectly solved, all reliable areas are expanded and eroded to obtain the CSI and BSI areas of A and B to roughly correspond to the shallow saturation area and the deep saturation area of ​​A and B. The CSI and BSI interpolation methods are used to interpolate and complete the CSI and BSI areas of A and B.

[0097] S4. Find the unsaturated lower intersection points between adjacent stripes. The surface where these points are located is F lower , using these lower intersection points as F lower reliable points, and use A, B and F lower The relationship between the two points is expanded by using A and B of the reliable point position. lower The reliable point, finally use F lower The extended reliable region is expanded and eroded to obtain F lower The CSI and BSI areas roughly correspond to F lower The shallow saturation area and deep saturation area are obtained by interpolating the CSI and BSI regions. lower Perform interpolation and completion;

[0098] S5. According to A, B and F lower The relationship between the three surfaces, A, B and F obtained by interpolation lower Calculate the exact A for each position;

[0099] S6. Using the relationship between stripes, the saturated stripes are repaired using the repaired A and the unsaturated stripes. After the saturated portion at the top of the stripe is repaired, the remaining portion is completed using CSI interpolation.

[0100] S7, when all fringes are repaired, use the multi-step phase shift method to calculate the phase;

[0101] S8. Use the complementary Gray code method to solve the order and unfold the wrapped phase.

[0102] Please refer to Figure 1 Learned:

[0103] Step a, projecting a grayscale sinusoidal fringe pattern onto a highly reflective surface, and collecting deformed saturated fringes with a CCD;

[0104] Step b, solving the reliable area and the parameters A and B of the reliable area;

[0105] Step c: Calculate the CSI and BSI areas of A and B by dilation and corrosion, and use the CSI-BSI interpolation method to interpolate and complete the CSI and BSI areas of A and B;

[0106] Step d. Find F lower Reliable point, expansion corrosion calculation F lower The CSI and BSI areas are interpolated using the CSI-BSI method to calculate the F lower Perform interpolation and completion;

[0107] Step e: Use interpolated A, B, and F lower Calculate the exact A for each position;

[0108] Step f: Use the repaired A and the unsaturated stripes to repair the saturated stripes, and use CSI interpolation to complete the rest;

[0109] Step g, calculating the wrapping phase using a multi-step phase shift method;

[0110] Step h: Expand the wrapped phase using the complementary Gray code method.

[0111] Second embodiment

[0112] The specific steps of step a are as follows:

[0113] Step a1: Encode N sinusoidal fringe patterns as a projection image group.

[0114] The phase-shift fringe encoding formula is:

[0115]

[0116] Where N is the number of phase shift steps, n = 1, 2, ..., N is the phase shift sequence number, f0 is the set fringe frequency, and a and b are generally set to 127.5.

[0117] The S1 includes the following specific steps:

[0118] S1, projecting a grayscale sinusoidal fringe pattern onto a highly reflective surface, the CCD collects the deformed saturated fringes, the projected N fringe patterns have a constant phase difference δ, and the captured image also has a phase difference of δ, that is,

[0119]

[0120] in, N is the number of phase shift steps, which is an integer greater than or equal to 3, n=1, 2, ..., N is the phase shift sequence, A(x, y) is the average intensity, and B(x, y) is the intensity modulation.

[0121] Third embodiment

[0122] The S2 includes the following specific steps:

[0123] S21. For each point in the image, divide it into a reliable area, a shallow saturation area, and a deep saturation area according to the saturation of the N phase shift images. If the values ​​of three or more of the N phase shift images at a certain point are less than 255, then this point is a reliable area, and the rest are unreliable areas. In the unreliable area, assume that A a is the actual A parameter, then when A a When A<255, the point is a shallow saturation area. a When ≥255, this point is in the deep saturation area.

[0124] S22. The A parameter of the reliable area can be solved directly. The A parameter of the shallow saturation area and the deep saturation area cannot be solved directly. The formula for solving the A parameter of the reliable area is as follows:

[0125] According to Euler's formula e ix =cosx+isinx, I n (x,y) can be written as:

[0126]

[0127] The stripe subscripts are n1,n2,...,n k The stripes are not saturated. Combine the equations for these unsaturated stripes at the (x,y) position:

[0128]

[0129] Then the formula for solving the A and B parameters is:

[0130]

[0131] Fourth embodiment

[0132] The S3 includes the following specific steps:

[0133] S31, after solving the reliable regions A and B, the reliable regions are expanded. The expansion distance should be more than half of the fringe period to obtain the expanded region;

[0134] Then the expanded area is eroded, and the erosion distance should be greater than the expansion distance to obtain the CSI area;

[0135] Finally, the CSI region is inverted to obtain the BSI region. Since the shallow saturation region and the deep saturation region cannot be perfectly solved, the CSI and BSI regions are obtained to roughly correspond to the shallow saturation region and the deep saturation region;

[0136] S32. Using the A values ​​of all reliable points in the CSI area as a basis, interpolate and complete the A value of the entire CSI area.

[0137] Interpolate and complete the B value of the entire CSI area based on the B values ​​of all reliable points in the CSI area;

[0138] The interpolation method uses the cubic spline interpolation method: cubic spline interpolation, referred to as CSI interpolation method;

[0139] S33, using the A values ​​of all reliable points in the BSI area as a basis, interpolating and completing the A value of the entire BSI area;

[0140] The B values ​​of all reliable points in the BSI area are used as the basis to interpolate and complete the B values ​​of the entire BSI area;

[0141] The interpolation method uses the biharmonic spline interpolation method: biharmonic spline interpolation, referred to as BSI interpolation method.

[0142] The S4 includes the following specific steps:

[0143] S41、F lower The method for finding reliable points is as follows: the reliable point is the lower intersection point of adjacent stripes without saturation and the lower reliable point calculated by the reliable area of ​​A and B. n (x,y) and I n+1 (x,y), the equation of the intersection point is:

[0144]

[0145] Solving the above equation, the phase of the intersection position is:

[0146]

[0147] The fringe intensity value at the intersection position is:

[0148]

[0149] k is an integer. In order to determine whether these intersection points are lower intersection points, use I n (x,y) and I n+1 The intensity of the intersection of the (x,y) symmetrical stripes at that position is used to assist in the determination;

[0150] I n (x,y) and I n+1 The symmetrical stripes of (x,y) are represented as, MOD(N / 2+n) and I MOD(N / 2+n+1), where 'MOD' borrows the concept of the remainder function and is expressed as:

[0151]

[0152] N represents the total number of phase shift steps, and the intensity of the symmetrical fringe intersection is expressed as:

[0153]

[0154] determination The point is the lower intersection point P lower int The conditions are:

[0155]

[0156]

[0157] All F lower All lower intersection points of are the union of all adjacent stripe intersection points:

[0158]

[0159] In addition, A and B at the reliable position can also be used to control F lower The reliable points are further expanded, and the expanded formula is:

[0160]

[0161] S42, solve to F lower After the reliable area is obtained, the reliable area is expanded, and the expansion distance should be more than half of the stripe period to obtain the expanded area;

[0162] Then the expanded area is eroded, and the erosion distance should be greater than the expansion distance to obtain the CSI area;

[0163] Finally, the CSI region is inverted to obtain the BSI region;

[0164] Since the shallow saturation region and the deep saturation region cannot be solved perfectly, the CSI and BSI regions are obtained to roughly correspond to the shallow saturation region and the deep saturation region;

[0165] S43, using F of all reliable points in the CSI area lower The value is based on the F of the entire CSI area lower The value is interpolated and completed;

[0166] The interpolation method uses the cubic spline interpolation method: cubic spline interpolation, referred to as CSI interpolation method;

[0167] S44, use F of all reliable points in the BSI area lower The value is based on the F of the entire BSI area lower The interpolation method uses the biharmonic spline interpolation method: biharmonic spline interpolation, referred to as BSI interpolation method.

[0168] The S5 includes the following specific steps:

[0169] S51, interpolated surfaces A, B and F lower In the figure, the strength of A and B is higher, the interpolation base points are few, and the error with the ideal surface is large. lower The interpolation base points are many and uniform, and the error with the ideal surface is small. The interpolated surfaces A, B, and F lower Written as A′, B′, F lo ' wer , the ideal surfaces are A, B, F lower , because the interpolation function can preserve the gradient when interpolating two sets of points with different intensities of the same trend, the ratio of A′, B′ is very close to the ratio of A, B:

[0170]

[0171] Therefore, the surfaces A″ and B″ with very small errors from the ideal surfaces A and B can be solved:

[0172]

[0173] The S6 includes the following specific steps:

[0174] S61, when an image is saturated in N-step phase shift, the fringe repair value I of the saturated area n r The repair formula is as follows:

[0175] When N is an even number:

[0176]

[0177] When N is an odd number:

[0178]

[0179] S62. After the saturated phase-shifted image is repaired using the above repair formula, if there are still saturated positions, these positions are interpolated and completed using the CSI interpolation method based on the global unsaturated fringe values ​​and the repaired fringe values.

[0180] The S7 includes the following specific steps:

[0181] S71. The wrapped phase can be solved by repairing N fringe patterns:

[0182]

[0183] The phase obtained by the above formula is the wrapped phase. To obtain height information, the unwrapped phase needs to be used. The absolute phase method can unwrap the wrapped phase.

[0184] The S8 includes the following specific steps:

[0185] S81, projecting multiple Gray code fringe patterns onto the surface of a high reflectivity object, and using a CCD to obtain Gray code fringe patterns. The first few patterns are used to calculate the order, and the last pattern is used to correct the order to obtain the order K of the fringe.

[0186] S82. Use the order K to unfold the wrapped phase. The unfolding formula is as follows:

[0187]

[0188] The final Φ(x,y) obtained is the unfolded phase.

[0189] Fifth embodiment

[0190] The specific steps of step d are as follows:

[0191] Step d1: Find the lower intersection point where adjacent stripes are not saturated. Take the four-step phase shift as an example, Figure 2 As shown, Figure 2 a shows the positions of all upper and lower intersection points in the four-step phase shift.

[0192] Figure 2 b illustrates the process of determining the upper and lower intersection points using I1 and I2 as examples. These points are used to retrieve values ​​on symmetrical stripes at the same position. The intersection points are represented by solid dots. The reference point is indicated by a hollow dot The intersection and reference points By comparing them, we can determine whether they are up or down. The intersection point is higher than the reference point, indicating an upper intersection point. On the contrary, when When the intersection point is below the reference point, it is indicated as a lower intersection point. However, it is important to emphasize that when the saturation level is high, may not exist, in which case if If it does not exist, then the intersection It should still be considered as the lower intersection point, and the above method is used to find the upper and lower intersection points of I1 and I2 respectively.

[0193] Figure 3This is the phase comparison diagram calculated before and after fringe repair. It can be seen that after fringe saturation repair, the calculated phase error is significantly reduced, the accuracy is improved, and no additional fringe projection and equipment assistance are required. The measurement speed is fast and the equipment is simple.

[0194] The high dynamic range fringe projection three-dimensional measurement method based on fringe restoration provided by the present invention has the following beneficial effects:

[0195] A high dynamic range fringe projection 3D measurement method based on fringe restoration can effectively restore saturated fringes and improve the dynamic range of the system. Using the fringe restoration method, the information utilization rate of each image is increased, the number of projection patterns required for high dynamic range measurement is reduced, and the measurement speed is effectively improved.

[0196] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A high dynamic range fringe projection three-dimensional measurement method based on fringe restoration, characterized in that: The following steps are involved: S1, projecting a multi-step grayscale sinusoidal fringe pattern onto a highly reflective surface, and the CCD captures the multi-step phase-shifted, deformed, saturated fringes; S2. Divide the captured image into reliable areas, shallow saturation areas, and deep saturation areas. Calculate the reliable areas using a multi-step phase-shift saturation fringe pattern. Calculate the A and B parameters of the reliable areas using the Euler formula method. S3. Since the shallow saturation region and the deep saturation region cannot be perfectly solved, all reliable regions are expanded and eroded to obtain the CSI and BSI regions of A and B corresponding to the shallow saturation region and the deep saturation region of A and B. The CSI and BSI interpolation methods are then used to interpolate and complete the CSI and BSI regions of A and B. S4. Find the unsaturated lower intersection points between adjacent stripes. The surface where these points are located is F lower , these lower intersection points are taken as F lower reliable points, and use A, B and F lower The relationship between the two points is expanded by using A and B of the reliable point position. lower The reliable point, finally use F lower The extended reliable region is expanded and eroded to obtain F lower The CSI and BSI areas correspond to F lower The shallow saturation area and deep saturation area are obtained by interpolating the CSI and BSI regions. lower Perform interpolation and completion; S5. According to A, B and F lower The relationship between the three surfaces, A, B and F obtained by interpolation lower Calculate the exact A for each position; S6. Using the relationship between stripes, the saturated stripes are repaired using the repaired A and the unsaturated stripes. After the saturated portion at the top of the stripe is repaired, the remaining portion is completed using CSI interpolation. S7, when all fringes are repaired, use the multi-step phase shift method to calculate the phase; S8, using the complementary Gray code method to solve the order and unfold the wrapped phase; The S1 includes the following specific steps: S1, projecting a grayscale sinusoidal fringe pattern onto a highly reflective surface, the CCD collects the deformed saturated fringes, the projected N fringe patterns have a constant phase difference δ, and the captured image also has a phase difference of δ, that is, in, N is the number of phase shift steps, which is an integer greater than or equal to 3, n = 1, 2, ..., N is the phase shift sequence, A (x, y) is the average intensity, B (x, y) is the intensity modulation, is the object phase; The S3 includes the following specific steps: S31, after solving the reliable regions A and B, the reliable regions are expanded. The expansion distance should be more than half of the fringe period to obtain the expanded region; Then the expanded area is eroded, and the erosion distance should be greater than the expansion distance to obtain the CSI area; Finally, the CSI region is inverted to obtain the BSI region. Since the shallow saturation region and the deep saturation region cannot be perfectly solved, the CSI and BSI regions are obtained to correspond to the shallow saturation region and the deep saturation region; S32. Using the A values ​​of all reliable points in the CSI area as a basis, interpolate and complete the A value of the entire CSI area. Interpolate and complete the B value of the entire CSI area based on the B values ​​of all reliable points in the CSI area; The interpolation method uses the cubic spline interpolation method: cubic spline interpolation, referred to as CSI interpolation method; S33, using the A values ​​of all reliable points in the BSI area as a basis, interpolating and completing the A value of the entire BSI area; The B values ​​of all reliable points in the BSI area are used as the basis to interpolate and complete the B values ​​of the entire BSI area; The interpolation method uses the biharmonic spline interpolation method: biharmonic spline interpolation, referred to as BSI interpolation method; The S4 includes the following specific steps: S41、F lower The method for finding reliable points is as follows: the reliable point is the lower intersection point of adjacent stripes without saturation and the lower reliable point calculated by the reliable area of ​​A and B. n (x,y) and I n+1 (x,y), the equation of the intersection point is: Solving the above equation, the phase of the intersection position is: The fringe intensity value at the intersection position is: k is an integer. In order to determine whether these intersection points are lower intersection points, use I n (x,y) and I n+1 The intensity of the intersection of the (x,y) symmetrical stripes at that position is used to assist in the determination; I n (x,y) and I n+1 The symmetrical stripes at (x,y) are represented as I MOD(N / 2+n) and I MOD(N / 2+n+1) , where 'MOD' borrows the concept of the remainder function and is expressed as: N represents the total number of phase shift steps, and the intensity of the symmetrical fringe intersection is expressed as: determination The point is the lower intersection point P lowerint The conditions are: is the if is the if All F lower All lower intersection points of are the union of all adjacent stripe intersection points: In addition, A and B at the reliable position can also be used to control F lower The reliable points are further expanded, and the expanded formula is: S42, solve to F lower After the reliable area is obtained, the reliable area is expanded, and the expansion distance should be more than half of the stripe period to obtain the expanded area; Then the expanded area is eroded, and the erosion distance should be greater than the expansion distance to obtain the CSI area; Finally, the CSI region is inverted to obtain the BSI region; Since the shallow saturation region and the deep saturation region cannot be solved perfectly, the CSI and BSI regions are obtained to correspond to the shallow saturation region and the deep saturation region; S43, using F of all reliable points in the CSI area lower The value is based on the F of the entire CSI area lower The value is interpolated and completed; The interpolation method uses the cubic spline interpolation method: cubic spline interpolation, referred to as CSI interpolation method; S44, use F of all reliable points in the BSI area lower The value is based on the F of the entire BSI area lower The interpolation method uses the biharmonic spline interpolation method: biharmonic spline interpolation, referred to as BSI interpolation method.

2. The high dynamic range fringe projection 3D measurement method based on fringe restoration according to claim 1, characterized in that: The S2 includes the following specific steps: S21. For each point in the image, divide it into reliable area, shallow saturation area and deep saturation area according to the saturation of N phase shift images. If the value of three or more of the N phase shift images at a certain point is less than 255, then this point is a reliable area, and the rest are unreliable areas. In the unreliable area, it is assumed that A a is the actual A parameter, then when A a When A is less than 255, the point is a shallow saturation area. a When ≥255, this point is the deep saturation area. S22. The A parameter of the reliable area is solved directly. The A parameter of the shallow saturation area and the deep saturation area cannot be solved directly. The A parameter solution formula of the reliable area is as follows: According to Euler's formula e ix =cosx+isinx, I n (x,y) is written as: The stripe subscripts are n1,n2,...,n k The stripes are not saturated. Combine the equations for these unsaturated stripes at the (x,y) position: Then the formula for solving the A and B parameters is: The real function represents the real part of a complex number.

3. The high dynamic range fringe projection 3D measurement method based on fringe restoration according to claim 1, characterized in that: The S5 includes the following specific steps: S51, interpolated surfaces A, B and F lower In the figure, the strength of A and B is higher, the interpolation base points are few, and the error with the ideal surface is large. lower The interpolation base points are many and uniform, and the error with the ideal surface is small. The interpolated surfaces A, B, and F lower Written as A′, B′, F′ lower , the ideal surfaces are A, B, F lower , because the interpolation function can preserve the gradient when interpolating two sets of points with different intensities of the same trend, the ratio of A′, B′ is very close to the ratio of A, B: Here k(x,y) is directly defined as the ratio of A'(x,y) to B'(x,y), which is very close to the ratio of A(x,y) to B(x,y); Therefore, the surfaces A″ and B″ with very small errors from the ideal surfaces A and B can be solved:

4. The high dynamic range fringe projection 3D measurement method based on fringe restoration according to claim 1, characterized in that: The S6 includes the following specific steps: S61, when an image is saturated in N-step phase shift, the fringe repair value of the saturated area The repair formula is as follows: When N is an even number: When N is an odd number: S62. After the saturated phase-shifted image is repaired using the above repair formula, if there are still saturated positions, these positions are interpolated and completed using the CSI interpolation method based on the global unsaturated fringe values ​​and the repaired fringe values.

5. The high dynamic range fringe projection 3D measurement method based on fringe restoration according to claim 1, characterized in that: The S7 includes the following specific steps: S71. Obtain the wrapped phase by reconstructing the N restored fringe patterns: The phase obtained by the above formula is the wrapped phase. To obtain height information, the unwrapped phase needs to be used. The absolute phase method unwraps the wrapped phase.

6. The high dynamic range fringe projection 3D measurement method based on fringe restoration according to claim 1, characterized in that: The S8 includes the following specific steps: S81, projecting multiple Gray code fringe patterns onto the surface of a high reflectivity object, and using a CCD to obtain Gray code fringe patterns. The first few patterns are used to calculate the order, and the last pattern is used to correct the order to obtain the order K of the fringe. S82. Use the order K to unfold the wrapped phase. The unfolding formula is as follows: The final Φ(x,y) obtained is the unfolded phase.

Images