A fast phase unwrapping method based on reference plane speckle assistance and application

Abstract

The application discloses a kind of based on reference plane speckle auxiliary fast phase unwrapping method and application, including the following steps: acquisition reference plane speckle diagram, and reference plane absolute phase is obtained by space phase unwrapping method;Three frames of phase shift fringe patterns and object speckle diagram of imaging object are collected, and object wrapped phase is solved according to three frames of phase shift fringe patterns of imaging object;Disparity map is generated based on the speckle texture features matching in object speckle diagram and reference plane speckle diagram;Phase unwrapping is carried out to object wrapped phase using reference plane absolute phase, and composite phase containing layered feature is generated;Using disparity map, the phase order of composite phase is recalibrated layer by layer, and the accurate absolute phase of imaging object is obtained.The application combines the advantages of traditional phase shift code and speckle code, and the calculation of absolute phase can be realized without additional auxiliary image, and the number of projection is significantly less than that of traditional method, which greatly improves the efficiency of three-dimensional imaging measurement.

Description

TECHNICAL FIELD

[0001] The application relates to a reference plane speckle auxiliary-based fast phase unwrapping method and application, and belongs to the technical field of three-dimensional shape measurement. BACKGROUND

[0002] In the face of growing demand for information richness and diversity, the research focus of computer vision is undergoing a significant expansion from two-dimensional planes to three-dimensional space. In this context, how to efficiently and accurately obtain spatial stereo vision perception information has become one of the core topics of current computer vision research.

[0003] Optical three-dimensional measurement technology has been widely used in robot navigation, industrial detection modeling, microscopic observation and other fields due to its low cost, high precision and non-contact advantages. Among them, the method based on structured light occupies the mainstream, and fringe projection profilometry (FPP) is a typical representative. This technology projects a grating fringe carrying specific phase information to the measured object through a projector, and a camera synchronously captures the deformed fringe image modulated by the object surface topography. The wrapped phase is demodulated by using a phase shift algorithm. Since the object depth information is encoded in the phase distribution, and the phase is calculated by the inverse tangent function, its value is limited in the principal value interval, which leads to the periodic ambiguity of the phase, i.e. the ambiguity problem. In order to eliminate ambiguity and obtain continuous absolute phase reflecting the real surface profile, phase unwrapping (PU) technology must be used.

[0004] It is a key step for three-dimensional reconstruction to realize robust unwrapping of wrapped phase to absolute phase. After long-term research, various phase unwrapping methods have been developed, mainly divided into spatial phase unwrapping method and temporal phase unwrapping method. The spatial phase unwrapping method does not need to project an additional auxiliary pattern, only relies on the phase continuity between the spatial neighborhood pixels to solve by path integration or global optimization, and has high calculation efficiency, such as a binocular grating projection measurement method based on spatial phase unwrapping disclosed in Chinese patent application No. CN201910812511.6, however, the unwrapping result is prone to error when the spatial phase unwrapping method faces the scene with depth mutation, isolated region or low signal-to-noise ratio region, and the error will spread along the unwrapping path. The temporal phase unwrapping method projects a series of auxiliary patterns with specific coding, such as Gray code, multi-frequency fringe, to the scene, and assigns a unique order identification to each wrapped phase period, so that the absolute phase order of each pixel can be determined independently and unambiguously in theory, completely avoiding the error propagation problem in the spatial phase unwrapping method, such as a high-precision three-dimensional face measurement method based on near-infrared light projection disclosed in Chinese patent application No. CN202110016306.6, but the temporal phase unwrapping method depends on the mode of multi-frame image acquisition, significantly limits the measurement speed, and has poor adaptability to dynamic scenes, and the noise or system error on the time axis may accumulate, resulting in artifacts. Therefore, it is a crucial goal in this field to develop a phase unwrapping method with high robustness, high precision and without additional mode (especially reducing the frame number dependence). SUMMARY

[0005] The technical problem solved by the present application is to provide a fast phase unwrapping method based on reference plane speckle assistance and application to solve the problems of large error and low efficiency of existing phase unwrapping methods.

[0006] The technical scheme adopted by the present application is as follows:

[0007] The present application first discloses a fast phase unwrapping method based on reference plane speckle assistance, comprising the following steps:

[0008] S1, project an initial pattern containing a fringe pattern and a speckle pattern on a reference plane, collect a reference plane speckle pattern, and obtain a reference plane absolute phase by a spatial phase unwrapping method;

[0009] S2, place an imaging object in front of the reference plane to project the initial pattern in step S1, and synchronously collect a deformed phase shift fringe pattern of the imaging object and an object speckle pattern, and calculate an object wrapped phase according to three phase shift fringe patterns of the imaging object;

[0010] S3, generate a disparity map by stereo matching based on the speckle texture features in the object speckle pattern and the reference plane speckle pattern;

[0011] S4, phase unwrapping the object wrapped phase by using the reference plane absolute phase to generate a composite phase containing layered features;

[0012] S5, recalibrating the phase order of the composite phase layer by layer by using the disparity map in step S3 to obtain the accurate absolute phase of the imaged object.

[0013] In the fast phase unwrapping method based on reference plane speckle assistance, further, in step S1, the initial pattern is projected onto the reference plane by designing a phase shift fringe pattern, a binary fringe pattern and a speckle pattern to construct a projection sequence, and a first deformed phase shift fringe pattern, a deformed binary fringe pattern and a reference plane speckle pattern are obtained by synchronously shooting with an industrial camera, the reference plane wrapped phase is obtained from the first deformed phase shift fringe pattern, the phase order corresponding to the reference plane wrapped phase is obtained according to the deformed binary fringe pattern of the reference plane, and then the absolute phase distribution of the reference plane is calculated.

[0014] In the fast phase unwrapping method based on reference plane speckle assistance, further, in step S2, the same initial pattern as in step S1 is used to project the imaged object, a second deformed phase shift fringe pattern of the imaged object and an object speckle pattern are synchronously shot by an industrial camera, and the second deformed phase shift fringe pattern is used to calculate the imaged object wrapped phase by using the phase shift method, and the formula is as follows:

[0015]

[0016] wherein, represents the imaged object wrapped phase, N is the number of fringe patterns, is the second deformed phase shift fringe pattern of the imaged object, represents the horizontal and vertical coordinates of the image plane of the industrial camera.

[0017] In the fast phase unwrapping method based on reference plane speckle assistance, further, step 3 includes the following steps:

[0018] Step S31, Census transformation is performed on the reference plane speckle pattern and the object speckle pattern to generate a binary descriptor, and a matching cost is calculated based on the binary descriptor;

[0019] Step S32, in the PatchMatch framework, the matching cost is used to optimize the disparity map between the reference plane speckle pattern and the object speckle pattern.

[0020] ​​In the fast phase unwrapping method based on reference plane speckle assistance according to the application, further, in the step S31, a Census transform is used to construct a local structure descriptor: all pixels in the reference plane speckle pattern and the object speckle pattern are sampled with a neighborhood window, and a Hamming distance between a Census descriptor Census(p) of a pixel point p in the object speckle pattern and a Census descriptor Census(q) of a candidate matching pixel point q in the reference plane speckle pattern is defined as a matching cost, which is expressed as follows:

[0021] ,

[0022] ,

[0023] ,

[0024] wherein, , are the i-th binary values of a 8-bit string and a 16-bit string respectively, represents a gray value of the pixel point p, is the i-th pixel of the window with the pixel point p as the center, represents the matching cost of the pixel point p and the pixel point q, represents an exclusive OR operation. In the fast phase unwrapping method based on reference plane speckle assistance according to the application, further, in the step S32,

[0025] firstly, taking the object speckle pattern as a reference image, a random initialization is performed on each pixel point p in the object speckle pattern as follows:

[0026]

[0027] ,

[0028] ,

[0029] wherein, represents an initial disparity value, represents a uniform distribution, and are preset disparity range boundary values, represents an initial cost, represents a corresponding point in the reference plane speckle pattern;

[0030] then, the pixels in the two speckle patterns are processed in a row scanning order, and the disparity value is iteratively optimized for K times, and a scanning direction alternates with the parity of the iteration number, and a spatial propagation, a view propagation and an adaptive random search are sequentially performed on each pixel point p by the following formula respectively:​​​

[0031] ,

[0032] where k represents the iteration number, , represents the disparity value of the pixel point p in the kth iteration, represents the matching cost of the pixel point p when the disparity is d, represents the set of neighborhood pixels processed in the current scanning direction, and if the disparity of the neighborhood pixel corresponds to a smaller cost, the disparity of the current pixel is updated, represents the inverse of the disparity value of the corresponding point in the reference plane speckle pattern relative to the object speckle pattern in the last iteration, , R is a random number uniformly distributed in the interval [-1, 1], represents the disparity range, is a decay factor,

[0033] Finally, after K iterations of all pixel disparity values, the final disparity map between the reference plane speckle pattern and the object speckle pattern is output satisfies the global energy minimization constraint.

[0034] In a fast phase unwrapping method based on reference plane speckle assistance according to the application, further, in step 4, based on the geometric position constraint of the imaged object and the reference plane, the phase order of the object wrapped phase is obtained using the absolute phase distribution of the reference plane, and the composite phase containing layered features is obtained by combining the object wrapped phase and the phase order.

[0035] In a fast phase unwrapping method based on reference plane speckle assistance according to the application, further, the phase order of the object wrapped phase is calculated by the following formula:

[0036] ,

[0037] where, is the upward rounding operator, represents the absolute phase of the reference plane, represents the object wrapped phase, represents the coordinate of the corresponding pixel;

[0038] The calculation formula of the composite phase is:

[0039] ,

[0040] where, represents the composite phase of the object containing layered features.

[0041] In the fast phase unwrapping method based on reference plane speckle assistance, further, the step S5 comprises the following sub-steps:

[0042] Step S51, the composite phase obtained in step S4 is traversed pixel by pixel, and the method of flood fill is used to divide the region layer by layer, and the mathematical expression of the divided region is as follows:

[0043] ,

[0044] Wherein, represents the flood fill operator, represents the composite phase of step S4, represents the zth region obtained by segmentation, represents the number of regions obtained by segmentation;

[0045] Step S52, using the parallax map obtained in step S3, the absolute phase of the imaging object at the corresponding position of the reference plane is found pixel by pixel by the following formula:

[0046] ,

[0047] Wherein, represents the absolute phase at the corresponding position of the reference plane obtained by parallax, represents the parallax value of the parallax map at ;

[0048] Step S53, the phase order of each layered region of the composite phase is recalculated by the following formula, and the accurate absolute phase of the imaging object is obtained:

[0049] ,

[0050] ,

[0051] ,

[0052] Wherein, represents the region where the pixel exists, represents the accurate absolute phase of the region , represents the composite phase of the region , represents the phase order difference between the accurate phase value and the composite phase of the region , and the average value is calculated according to the region, represents the rounding operation, represents the output accurate absolute phase of the imaging object.

[0053] The present invention also discloses a three-dimensional imaging method, which uses the fringe projection profile measurement method for three-dimensional imaging. During the imaging process, the present invention employs a fast phase unfolding method based on reference plane speckle assistance to unfold the projected fringe pattern.

[0054] The present invention, by adopting the above technical solution, has the following beneficial effects:

[0055] 1. This invention employs monocular stereo matching to match a reference plane and the object under test from the same viewpoint. The absolute phase of the object is then obtained by mapping the reference plane. Through the synergistic optimization of the geometric constraint expansion method of reference plane phase-wrapped phase and the binary speckle stereo matching algorithm, the measurement accuracy and efficiency are both improved. This solves the problems of insufficient imaging efficiency and motion temporal error accumulation in traditional multi-frame projection schemes.

[0056] 2. This invention only requires a three-frame phase shift method that meets the minimum projection frame requirement. It innovatively integrates a speckle-assisted phase unfolding mechanism to reduce the number of projection frames. Especially in dynamic scenes, it not only improves the speed but also reduces the error accumulation problem caused by motion. Robust unfolding can be achieved without additional projection of the coded pattern, which greatly improves the encoding and decoding efficiency.

[0057] 3. Compared with the geometrically constrained phase unfolding method, this invention introduces speckle matching to remove the object thickness limitation, breaks through the object thickness limitation, significantly suppresses ambient light and noise interference, significantly improves robustness, and can realize high-precision and rapid reconstruction of complex three-dimensional surfaces, enabling high-precision rapid measurement of complex three-dimensional surfaces.

[0058] In summary, the fast phase unfolding method based on reference plane speckle assistance provided by this invention combines the advantages of traditional phase shift codes and speckle codes. It can realize the calculation of absolute phase without additional auxiliary images, and the number of projections is significantly less than that of traditional methods, which greatly improves the efficiency of three-dimensional imaging measurement.

[0059] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of a fast phase unfolding method based on reference plane speckle assistance according to the present invention.

[0061] Figure 2a , 2b 2c and 2c represent the phase-shifted fringe pattern, binary fringe pattern, and speckle pattern in the initial projection pattern of the embodiment, respectively.

[0062] Figure 3a and 3b These are the reference plane speckle pattern and the object speckle pattern collected in the embodiment.

[0063] Figure 4a 、 4b and 4c are three first deformed phase-shifted fringe patterns of the reference plane captured in the example.

[0064] Figure 5 are deformed binary fringe patterns of the reference plane captured in the example.

[0065] Figure 6 are wrapped phase patterns of the reference plane obtained in the example.

[0066] Figure 7 are absolute phase patterns of the reference plane obtained in the example.

[0067] Figure 8a 、 8b and 8c are three second deformed phase-shifted fringe patterns of the imaged object captured in the example.

[0068] Figure 9 are wrapped phase patterns of the imaged object obtained in the example.

[0069] Figure 10 is the disparity map obtained in the example.

[0070] Figure 11 is the complex phase pattern containing layered features obtained in the example.

[0071] Figure 12 is the refined absolute phase pattern of the object obtained in the example. DETAILED DESCRIPTION

[0072] EMBODIMENT

[0073] The application will be further described below in connection with the drawings and embodiments. However, this should not be understood as limiting the scope of the above-mentioned subject matter of the application to the following embodiments, and any technology realized based on the content of the application falls within the scope of the application.

[0074] Referring to Figure 1 , a specific embodiment of the fast phase unwrapping method of the application is shown, which uses binary speckle particles with a size of 2x2 pixels and sinusoidal fringes with a wavelength of 25 as the encoding scheme, and uses a three-step phase shift method to measure a complex scene containing a mannequin. Specifically, the following steps are included:

[0075] S1, project an initial pattern containing a fringe pattern and a speckle pattern on a reference plane, capture a reference plane speckle pattern, and obtain a reference plane absolute phase by a spatial phase unwrapping method.

[0076] Specifically, the initial pattern is constructed by designing a phase-shift fringe pattern, a binary fringe pattern, and a speckle pattern to construct a projection sequence and project it onto a reference plane. A first deformed phase-shift fringe pattern, a deformed binary fringe pattern, and a speckle pattern of the reference plane are obtained by synchronously capturing images with an industrial camera. The phase wrapped by the reference plane is obtained from the first deformed phase-shift fringe pattern, and the phase order corresponding to the phase wrapped by the reference plane is obtained from the deformed binary fringe pattern. Then, the absolute phase distribution of the reference plane is calculated.

[0077] The formulas for generating the phase-shifted fringes and binary fringes of the initial pattern are as follows:

[0078] ,

[0079] .

[0080] in, Represents the x and y coordinates of the projection plane. Representing phase-shifted fringe patterns, such as Figure 2a As shown, For the average strength of the code, For encoding modulation intensity, For the fringe frequency, width represents the width of the projection plane. Representing a binary striped pattern, such as Figure 2b As shown.

[0081] Binary speckle pattern generated from a black low-resolution pattern ,like Figure 2c As shown, the formula for generating a binary speckle pattern is as follows:

[0082] ,

[0083] .

[0084] Where 1 represents white, 0 represents black, R represents a speckle image with a length of Q, and Q defines the physical size of the speckle (Q×Q pixels). As a uniform random variable, the speckle distribution plane is divided into 3×3 grid cells, with each cell retaining only one candidate point, thus ensuring the speckle density. The Chebyshev distance ensures that there are no adjacent speckles within the neighborhood of the speckle distribution.

[0085] Next, a phase-shifted fringe pattern, a binary fringe pattern, and a speckle pattern are projected onto the reference plane by a projector, and the speckle pattern on the reference plane, captured simultaneously by an industrial camera, is denoted as follows: ,like Figure 3a As shown. The formulas for the deformed phase-shift fringe pattern and the deformed binary fringe pattern are as follows:

[0086] 、

[0087] .

[0088] where the superscript represents the reference plane scene, denotes the image plane horizontal and vertical coordinates, denotes the first deformed phase-shifted fringe pattern of the reference plane scene captured by the camera, as shown in Figure 4a 、 4b and 4c, is the background intensity of the reference plane scene, is the mixed modulation intensity of the reference plane scene. is the absolute phase of the reference plane, denotes the deformed binary fringe pattern of the reference plane captured by the camera, as shown in Figure 5 , specifically denotes the gray value of the pixel point in the deformed binary fringe pattern of the reference plane.

[0089] The wrapped phase of the reference plane is obtained by the phase-shifting method, and the calculation formula is:

[0090] .

[0091] where denotes the wrapped phase of the reference plane, as shown in Figure 6 , due to the effect of function, the value range is wrapped between and , which is different from the absolute phase of the reference plane by an integer multiple of , which is called the phase order . The calculation of the phase order needs the help of the deformed binary fringe pattern of the reference plane, and the calculation formula is as follows:

[0092] .

[0093] where is a constant, whose value is between , the purpose is to find the center position of the deformed fringe after projection, denotes the floor function, specifically denotes the gray value of the pixel point in the deformed binary fringe pattern of the reference plane.

[0094] The pattern of the absolute phase of the reference plane is shown in Figure 7 , and the calculation formula is as follows:

[0095] .

[0096] S2, place the imaging object in front of the reference plane to project the initial pattern in step S1, and synchronously collect three frames of phase-shifted fringe patterns and object speckle patterns of the imaging object, and calculate the wrapped phase of the object according to the three frames of phase-shifted fringe patterns of the imaging object.

[0097] Place the imaging object in front of the reference plane, and synchronously capture the object speckle pattern obtained by the industrial camera, denoted as As shown in Figure 3b , the second deformed phase-shifted fringe pattern of the imaging object is denoted as:

[0098] .

[0099] wherein the superscript represents a background intensity containing a measurement scene of the imaging object, represents horizontal and vertical coordinates of an image plane, represents a camera-captured deformed phase-shifted fringe pattern of the object, and three frames of the second deformed phase-shifted fringe patterns of the imaging object are collected as shown in Figure 8a , 8b and 8c, is a background intensity containing a measurement scene of the imaging object, is a mixed modulation intensity containing a measurement scene of the imaging object. is an absolute phase of the object.

[0100] Then, the wrapped phase of the imaging object is obtained by the phase-shift method, and the calculation formula is:

[0101] .

[0102] wherein, represents the wrapped phase of the imaging object, and the pattern is shown in Figure 9 Due to the effect of the function, the value range is wrapped between .

[0103] S3, generate a disparity map by stereo matching based on speckle texture features in the object speckle pattern and the reference plane speckle pattern. Specifically, it includes the following two sub-steps:

[0104] Step S31, perform Census transformation on the reference plane speckle pattern and the object speckle pattern to generate binary descriptors, and calculate matching costs based on the binary descriptors.

[0105] Firstly, a local structure descriptor is constructed by Census transform: all pixels in the reference plane speckle pattern and the object speckle pattern are sampled by a neighborhood window, and their relative gray level relationship is coded as a 64-bit binary descriptor. The Hamming distance between the Census descriptor of a pixel p in the object speckle pattern and the Census descriptor of a candidate matching pixel q in the reference plane speckle pattern is defined as the matching cost, which is expressed as follows:

[0106]

[0107]

[0108]

[0109] wherein, is the i-th binary value of the 64-bit string, is the i-th binary value of the 64-bit string, is the gray value of the pixel p, is the i-th pixel of the window centered at the pixel p, is the matching cost of the pixel p and the pixel q, is the exclusive or operation. Step S32, under the PatchMatch framework, the disparity map between the reference plane speckle pattern and the object speckle pattern is optimized by using the matching cost.

[0110] Random initialization. Taking the object speckle pattern as the reference image, each pixel point in the image is executed as follows:

[0111]

[0112]

[0113]

[0114] wherein, is the initial disparity value, which is randomly generated according to a uniform distribution, is the uniform distribution, and are preset disparity range boundary values, is the initial cost, is the corresponding point in the reference speckle pattern.

[0115] Iterative propagation. The following is performed for iterations (a typical value is ), the i-th iteration (i ​​​​​​​​​) in a row-wise scan order, the scanning direction alternates with the parity of the iteration number, and for each pixel The spatial propagation, view propagation and adaptive random search are sequentially performed:

[0116] .

[0117] wherein k represents the iteration number, , represents the disparity value of the pixel point p in the kth iteration, represents the matching cost of the pixel point p when the disparity is d, represents a set of processed neighboring pixels in the current scanning direction, Taking the scanning from left to right and from top to bottom as an example, if the disparity corresponding to the cost of the neighboring pixel in the set of processed neighboring pixels is smaller, the disparity of the current pixel is updated, represents the inverse of the disparity value of the corresponding point in the reference plane speckle pattern relative to the object speckle pattern in the last iteration, , R is a random number uniformly distributed in the interval [-1, 1], represents the disparity range, , is a decay factor, usually taken as 0.5.

[0118] The disparity is output. After K iterations of the disparity values of all pixels, the disparity map between the final reference plane speckle pattern and the object speckle pattern is output satisfies the global energy minimization constraint, and the disparity map output by the embodiment is as shown in Figure 10 .

[0119] S4, using the absolute phase of the reference plane to unwrap the object wrapped phase to generate a composite phase containing hierarchical features. Specifically, based on the geometric position constraint of the imaged object and the reference plane, the phase order of the object wrapped phase is obtained using the absolute phase distribution of the reference plane, and the composite phase containing hierarchical features is obtained by combining the object wrapped phase with the phase order.

[0120] The calculation formula for obtaining the phase order of the object wrapped phase using the absolute phase distribution of the reference plane is:

[0121] .

[0122] wherein is a ceiling operator, represents the absolute phase of the reference plane, represents the object wrapped phase, represents the coordinate of the corresponding pixel.

[0123] Then, the object wrapping is added to the phase order to obtain a composite phase with layered features, calculated as follows:

[0124] .

[0125] in, This represents the composite phase of an imaged object containing layered features, as shown in the pattern. Figure 11 As shown, the composite phase is the composite absolute phase of the imaging object at this time. However, when the measured object exceeds the spatial range of the geometric constraints, the phase order changes. The calculation will be incorrect, resulting in an absolute phase error. It exhibits a hierarchical structure and lacks global uniqueness.

[0126] S5. Using the disparity map from step S3, perform layer-by-layer phase recalibration on the composite phase to obtain the accurate absolute phase of the imaged object. This includes the following sub-steps:

[0127] Step S51: Perform pixel-by-pixel traversal on the composite phase with layered features obtained in step S4, and use the flood filling method to divide the region into layers. The mathematical representation of the region division is as follows:

[0128] .

[0129] in, This represents the flood filling operator. This indicates the composite phase in step S4. This represents the z-th region obtained from the segmentation. Indicates the number of regions obtained from the segmentation;

[0130] Step S52: Using the disparity map obtained in step S3, find the absolute phase of the imaging object at the corresponding position on the reference plane pixel by pixel using the following formula:

[0131] .

[0132] in, This indicates that the absolute phase at the corresponding position in the reference plane is obtained through parallax. Indicates the disparity map in The disparity value at that location. Due to the presence of noise and matching error. It cannot be directly used as the absolute phase value of an object.

[0133] Step S53: Recalculate the phase order of each layered region of the composite phase using the following formula to obtain the accurate absolute phase of the imaged object:

[0134] ,

[0135] ,

[0136] .

[0137] wherein, represents the pixel wherein, represents the region of the accurate absolute phase, represents the complex phase of the region, represents the phase order difference between the accurate phase value and the complex phase of the region, and the average value is calculated according to the region, represents the rounding operation, represents the output accurate absolute phase of the imaged object, and the absolute phase pattern of the imaged object after unwrapping obtained by the embodiment is shown in Figure 12 .

[0138] The above-mentioned fast phase unwrapping method based on reference plane speckle assistance of the present application is applied to a three-dimensional imaging method, in particular, a three-dimensional imaging method using a fringe projection profilometry method. In the imaging process, the wrapped phase unwrapping of the projected fringe pattern is performed by using the reference plane speckle assistance phase unwrapping method. The calculation of the absolute phase of the imaged object can be realized without additional auxiliary images. The number of projections is significantly less than that of the traditional method, and the three-dimensional imaging measurement efficiency is greatly improved.

[0139] In this document, the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inner", "outer", "vertical", "horizontal", and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the purpose of expressing the technical solution clearly and conveniently, and therefore cannot be understood as a limitation on the present application.

[0140] In this document, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, in addition to including the listed elements, other elements not explicitly listed can also be included.

[0141] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A fast phase unwrapping method based on reference plane speckle assistance, characterized in that: The method comprises the following steps: S1, projecting an initial pattern containing a fringe pattern and a speckle pattern on a reference plane, collecting a reference plane speckle pattern, and obtaining a reference plane absolute phase by a spatial phase unwrapping method; S2, projecting the initial pattern in step S1 in front of an imaging object on the reference plane, and synchronously collecting a deformed phase shift fringe pattern of the imaging object and an object speckle pattern, and calculating an object wrapped phase according to three frame phase shift fringe patterns of the imaging object; S3, generating a disparity map based on speckle texture features in the object speckle pattern and the reference plane speckle pattern through stereo matching; S4, unwrapping the object wrapped phase using the reference plane absolute phase to generate a composite phase containing layered features; S5, recalibrating the phase order of the composite phase layer by layer using the disparity map in step S3 to obtain an accurate absolute phase of the imaging object, specifically comprising the following sub-steps: Step S51, traversing the composite phase containing layered features obtained in step S4 pixel by pixel, and dividing the regions layer by layer using a flood fill method, and the mathematical representation of the divided regions is as follows: ， wherein, denotes a flood fill operator, denotes a composite phase of step S4, denotes the z-th region obtained by the segmentation, denotes the number of regions obtained by the segmentation; Step S52, using the disparity map obtained in step S3 to find the absolute phase of the corresponding position of the imaging object on the reference plane pixel by pixel according to the following formula: ， wherein, denotes the absolute phase at the corresponding position of the reference plane by the parallax, denotes the absolute phase of the reference plane, denotes the parallax value of the parallax map at place; Step S53, recalculating the phase order of each layered region of the composite phase according to the following formula to obtain the accurate absolute phase of the imaging object: 、 、 ， wherein, represents a pixel the area in which represents an area the exact absolute phase of represents the complex phase of the area, represents the phase order difference between the exact phase value and the complex phase of the area, averaged by area, represents a rounding operation, represents the exact absolute phase of the imaged object output, is a ceiling operator.

2. The method of claim 1, wherein: In the step S1, the initial pattern is projected on the reference plane by designing a phase shift fringe pattern, a binary fringe pattern and a speckle pattern to construct a projection sequence, and a first deformed phase shift fringe pattern, a deformed binary fringe pattern and a reference plane speckle pattern are synchronously photographed by an industrial camera to obtain a reference plane wrapped phase, a phase order corresponding to the reference plane wrapped phase is obtained according to the deformed binary fringe pattern of the reference plane, and then the absolute phase distribution of the reference plane is calculated.

3. The method of claim 1, wherein: In the step S2, the same initial pattern as in step S1 is used to project the imaging object, and a second deformed phase shift fringe pattern of the imaging object and an object speckle pattern are synchronously photographed by an industrial camera to obtain an imaging object wrapped phase by using a phase shift method, and the formula is as follows: ， wherein, represents the wrapped phase of the imaged object, N is the number of fringe patterns, , is a second deformed phase-shifted fringe pattern of the imaged object, represents the image plane horizontal and vertical coordinates of the industrial camera.

4. The method of claim 1, wherein: The step 3 comprises the following steps: Step S31, performing Census transformation on the reference plane speckle pattern and the object speckle pattern to generate a binary descriptor, and calculating a matching cost based on the binary descriptor; Step S32, optimizing the disparity map between the reference plane speckle pattern and the object speckle pattern under the PatchMatch framework using the matching cost.

5. The method of claim 4, wherein: In the step S31, a local structure descriptor is constructed by Census transformation: all pixels in the reference plane speckle pattern and the object speckle pattern are sampled with a neighborhood window, and the Hamming distance between the Census descriptor Census(p) of a pixel point p in the object speckle pattern and the Census descriptor Census(q) of a candidate matching pixel point q in the reference plane speckle pattern is defined as the matching cost, and the formula is as follows: 、 、 ， in, , They are respectively Bit string and The i-th binary value of the bit string This represents the grayscale value of pixel p. Let p be the i-th pixel of the window centered at pixel p. This represents the matching cost between pixel p and pixel q. This represents the XOR operation.

6. The method of claim 5, wherein: In the step S32, First, take the object speckle pattern as the reference image, and take each pixel point in the object speckle pattern as the target pixel point The following random initialization is performed: 、 ， wherein, denotes an initial disparity value, denotes a uniform distribution, and is a pre-set disparity range boundary value, denotes an initial cost, denotes a corresponding point in the reference plane speckle pattern; Then the pixels in the two speckle patterns are processed in row scanning order, and the disparity value is iteratively optimized for K times, and the scanning direction is alternated according to the parity of the iteration number, and the spatial propagation, view propagation and adaptive random search are sequentially performed on each pixel point p by the following formula respectively: ， where k represents the iteration number, , represents the disparity value of pixel point p in the kth iteration, represents the matching cost of pixel point p when the disparity is d, represents the set of neighborhood pixels that have been processed in the current scanning direction, and if the disparity of the neighborhood pixel corresponds to a smaller cost, the disparity of the current pixel is updated, represents the inverse of the disparity value of the corresponding point in the reference plane speckle map relative to the object speckle map in the last iteration, , R is a random number uniformly distributed in the interval [-1, 1], represents the disparity range, is a decay factor, Finally, after K iterations of all pixel disparity values, output the final disparity map between the reference plane speckle pattern and the object speckle pattern satisfy global energy minimization constraints.

7. The method of claim 1, wherein: In step 4, based on the geometric position constraint of the imaging object and the reference plane, the phase order of the object wrapped phase is obtained using the absolute phase distribution of the reference plane, and the complex phase containing the layered feature is obtained by combining the object wrapped phase and the phase order.

8. The method of claim 7, wherein: The phase order of the object wrapped phase is calculated by the following formula: ， wherein, is a ceiling operator, denotes the absolute phase of the reference plane, denotes the object wrapped phase, denotes the coordinates the phase order of the corresponding pixel; The calculation formula of the complex phase is: ， wherein, represents a complex phase of the object containing the layered feature.

9. A method of three-dimensional imaging, characterized by: The fringe projection profilometry method is used for three-dimensional imaging, and a fast phase unwrapping method based on a reference plane speckle auxiliary in any one of claims 1-8 is used for wrapped phase unwrapping of the projected fringe pattern in the imaging process.

Images