PMD three-dimensional measurement method of adaptive fringe projection mode

Abstract

The invention discloses a PMD three-dimensional measurement method of a self-adaptive fringe projection mode. The method comprises the following steps: firstly, based on the surface topography characteristics of a measured mirror surface element, acquiring phase gradient distribution of the measured mirror surface element under a camera visual angle; secondly, according to the phase gradient distribution, self-adaptively modulating the local period of the projected fringes, so that the structured light fringes formed on the imaging surface of the camera after being reflected by the measured surface are integrally kept uniform and have moderate sparsity; projection is carried out by using the adaptive fringe pattern, a fringe image modulated by a measured surface is collected, and a corresponding relation between camera pixels and projection pixels is established; and finally, reconstructing the three-dimensional shape of the measured mirror surface element through ray tracing. According to the method, the local density of the stripes on the projection screen is adaptively adjusted according to the angle transformation characteristic of the reflected light of the measured surface, so that the measured structured light stripes in the visual angle of the camera are uniformly distributed and are properly sparse, and the PMD measurement precision is maximized under the condition that the morphology analysis integrity is ensured.

Description

Technical Field

[0001] This invention belongs to the technical field of high reflectivity surface identification and detection, and mainly relates to a PMD three-dimensional measurement method with adaptive stripe projection mode. Background Technology

[0002] Mirror finishing technology is the most effective means to improve the machining accuracy, fitting accuracy and service life of mechanical parts. Furthermore, it can also make the machined surface exhibit special optical properties and become an optical surface, which has wide applications in aerospace, automotive industry, semiconductor manufacturing, laser processing and other fields.

[0003] Surfaces formed using mirror finishing technology behave differently from ordinary machined surfaces. Light incident on these surfaces is completely mirror-reflected, making it impossible to accurately measure their three-dimensional shape by capturing texture information such as tool marks. Therefore, how to perform rapid, high-precision, non-contact optical measurements on mirror surfaces remains a challenge in the field of optical measurement.

[0004] Existing methods mainly fall into two categories: optical interferometry and phase measurement deflection (PMD). Optical interferometry obtains the morphological information of the measured surface with high precision by interfering the light beam reflected from the measured surface and the light beam reflected from the reference surface to form interference fringes. Since the reference surface is usually a standard plane or sphere, when measuring non-standard measured surfaces, interferometry techniques, which measure the difference between the standard surface and the measured surface at the wavelength level, exhibit significant limitations in dynamic range and adaptability, and the measurement device is complex. Phase deflection is a non-contact measurement method based on structured light. It uses a camera to capture the structured light fringe pattern projected onto a screen after being reflected and modulated by the measured surface, and then analyzes the pattern to reconstruct the three-dimensional morphology of the measured component surface. Compared with interferometry, PMD has advantages such as a large dynamic range, fast measurement speed, simple structure, and in-situ measurement, and is widely used in the three-dimensional morphology measurement of mirror components such as mirrors, transparent plates, deformable mirrors, and polished metal workpieces.

[0005] Conventional PMD technology measures the surface shape of a mirror by displaying stripes of equal period in different directions on a projection screen. However, when the surface being measured has a large local curvature, the structured light pattern captured by the camera may exhibit local stripe density that is either too sparse or too dense. This makes it impossible to guarantee that the global stripe density is within the optimal measurement range, resulting in reduced phase demodulation accuracy in the corresponding area or even demodulation errors, thus hindering the accuracy of the measurement results.

[0006] To overcome this challenge, researchers have recently explored PMD (partial surface depression) measurement of large local slope mirror elements by employing methods such as curved screens, wavefront conversion, source closure (projector array / multi-source stitching), and sub-aperture stitching. For example, Liu et al. replaced the planar screen with a curved screen and then unfolded the curved screen into a large-area, virtual planar screen to expand the measurement range of curved surface gradients; Blazer et al. introduced a cavlectometry method, which uses multiple screens or projector arrays to form a source closure surrounding the element under test, successfully measuring large local slope surfaces using the principle of reverse ray tracing; Graves et al. proposed an infinite deflectometry method, which fixes the element under test on a precision rotating stage and continuously rotates the element under test to achieve large local slope surface measurement through sub-aperture stitching; another method is multi-field stitching, which uses multiple cameras surrounding the element under test and stitches the surface shape data measured by each camera to measure the morphology of large gradient elements.

[0007] All of these PMD methods improve upon the previous ones by adding external components, such as screens or cameras. By adding measurement information in different measurement directions, they enhance the ability to solve surface shapes in areas with large curvature. Each method has its own measurement advantages and can be applied to different measurement objects. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention proposes a PMD three-dimensional measurement method based on an adaptive fringe projection mode.

[0009] This invention includes the following steps:

[0010] a) Based on the surface morphology features of the mirror element under test, obtain its phase gradient distribution under the camera's view.

[0011] b) Based on the phase gradient distribution, the local period of the projected stripes is adaptively modulated so that the structured light stripes formed on the camera imaging surface after reflection from the measured surface remain uniform and moderately sparse.

[0012] c) Project an adaptive fringe pattern and acquire a fringe image modulated by the surface under test to establish the correspondence between camera pixels and projection pixels;

[0013] d) Based on the correspondence, the three-dimensional shape of the tested mirror element is reconstructed by ray tracing.

[0014] Preferably, the phase gradient distribution is obtained by first projecting a set of equal-period horizontal and vertical large-period stripes and then analyzing the images acquired by the camera.

[0015] Preferably, the adaptive modulation step includes: mapping the phase gradient of the camera plane back to the display plane, and sparsening or densifying the stripe period of each region in the display plane using a preset adaptive threshold.

[0016] Preferably, mapping the phase gradient of the camera plane back to the display plane specifically involves: using the calibrated camera-screen geometry, inversely mapping the coordinates of each pixel on the camera imaging plane to the pixel coordinates on the display plane. In this way, each pixel on the display plane corresponds to a phase gradient value measured from the camera's viewpoint.

[0017] Preferably, the process of sparsening or densifying the stripe period of each region within the display plane using a preset adaptive threshold specifically involves:

[0018] A. Determine the density of local fringes: Set a threshold for the phase gradient and compare the gradient value at each location with this threshold.

[0019] If the gradient value is lower than the threshold, it means that the stripes in the corresponding region are too sparse in the camera's view.

[0020] If the gradient value is higher than the threshold, it means that the stripes are too dense.

[0021] B. Calculate the new local period:

[0022] In overly sparse regions, the fringe period is reduced; in overly dense regions, the fringe period is increased.

[0023] The new local period is inversely proportional to the measured phase gradient; the larger the gradient, the larger the period; the smaller the gradient, the smaller the period, thus ensuring that the fringe density in all regions of the camera tends to be consistent.

[0024] C. Generating a new phase distribution: The adjusted period is converted into a phase gradient field in the display plane, and then integrated along the horizontal and vertical directions to obtain the phase value corresponding to each pixel of the display. This phase distribution is the phase skeleton of the adaptive stripes.

[0025] Preferably, the projection using an adaptive stripe pattern specifically involves: modulating the phase skeleton into a sinusoidal grayscale stripe pattern, generating one in the horizontal and one in the vertical directions, and projecting them sequentially onto the mirror surface under test.

[0026] Preferably, the establishment of the correspondence between camera pixels and projection pixels introduces an intermediate phase plane, and accurate mapping is achieved through a lookup table between the phase plane and the display plane.

[0027] Preferably, the ray tracing is based on a dual-camera stereo vision model, and the three-dimensional coordinates of each point on the measured surface are calculated by combining camera-screen calibration parameters.

[0028] Preferably, the mirror element under test is an optical surface with large local curvature or a complex freeform surface.

[0029] Preferably, the adaptive fringe pattern can be dynamically refreshed during a single measurement to iteratively optimize the fringe density and further improve the phase resolution accuracy.

[0030] This invention addresses the challenge of conventional phase-resolution methods (PMDs) in simultaneously ensuring good analytical integrity and high measurement accuracy when measuring mirror elements with large surface gradients. The method utilizes the surface morphology characteristics of the measured mirror element to generate adaptive fringes with localized periodic variations. This ensures the uniformity of the structured light fringe distribution within the camera's field of view, thereby improving phase resolution accuracy, extending the dynamic range of the PMD measurement, and enhancing overall measurement precision. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the measurement principle of PMD;

[0032] Figure 2 This is a projection pattern diagram of vertical equal stripes;

[0033] Figure 3 An adaptive stripe pattern diagram;

[0034] Figure 4 This is a flowchart of the measurement process for PMD based on adaptive stripes.

[0035] Figure 5 A flowchart for constructing the correspondence between camera screen pixels and display plane pixels. Detailed Implementation

[0036] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0037] like Figure 1 The diagram shows the principle of adaptive fringe PMD measurement. The structured light fringe pattern projected by the display is reflected by the SUT and simultaneously captured by two cameras on either side of the display. By processing the fringe patterns captured by the cameras, the correspondence between each pixel on the camera and each pixel on the screen after the SUT's influence is obtained. Then, the three-dimensional shape of the SUT is obtained using ray tracing. These represent the position coordinate system and world coordinate system of camera 1, camera 2, and the screen, respectively.

[0038] like Figure 2As shown, the traditional PMD measurement method projects a periodic stripe pattern on the display. After reflection by the SUT, the resulting image on the camera is a stripe pattern with unequal intervals after being modulated by the SUT. This can lead to situations where the stripe pattern is too sparse or too dense in some areas, affecting the measurement accuracy and the completeness of the morphology.

[0039] like Figure 3 As shown, in order to avoid the appearance of excessively sparse or dense stripes, this application adaptively adjusts and optimizes the local density of the structured light stripes projected on the display, and finally obtains a relatively uniform structured light stripe pattern on the camera, which greatly improves the measurement accuracy and the solvability of the morphology.

[0040] In order to obtain Figure 3 To illustrate the technical effects, this application provides a flowchart of the measurement process for adaptive striped PMD, as shown in the embodiments. Figure 4 As shown:

[0041] Step 1: Calibrate the camera's in-camera screen coordinate system:

[0042] The intrinsic and extrinsic parameters of the two cameras are calibrated, and a precise transformation relationship between the camera coordinate system and the monitor screen coordinate system is established.

[0043] Step 2: Synchronously acquire the initial stripe image:

[0044] The image of the standard periodic (equally spaced) stripe pattern projected by the monitor and reflected by the mirror under test is recorded simultaneously by two cameras.

[0045] Step 3, Phase Extraction and Unfolding:

[0046] Phase extraction (usually using the phase shift method) is performed on the acquired fringe pattern, and phase unrolling is performed to obtain the absolute phase distribution at the camera pixel level.

[0047] Step 4: Calculate the phase gradient:

[0048] The phase gradient in the horizontal and vertical directions is calculated pixel by pixel on the camera image to measure the density of local stripes.

[0049] Step 5: Construct the phase gradient field of the screen plane:

[0050] Based on the phase gradient measured by the camera, the target phase gradient distribution required for each pixel position on the display screen is established through reverse mapping; the stripe density needs to be reduced at the screen end for areas with large gradients, and increased for areas with small gradients.

[0051] Step 6: Integrate to obtain the screen phase distribution:

[0052] The target phase gradient field obtained in step 5 is integrated in two dimensions to generate an absolute phase distribution on the screen plane that matches the curvature of the mirror.

[0053] Step 7: Modulate and generate an adaptive fringe pattern

[0054] The absolute phase from step 6 is sinusoidally modulated to generate a new adaptive fringe pattern, which is then projected back onto the mirror surface under test.

[0055] Step 8: Collect data again using dual cameras.

[0056] The same dual-camera system was used to record adaptive stripe images after specular reflection.

[0057] Step 9: Secondary Phase Extraction and Unfolding

[0058] Repeat the phase extraction and unfolding process; the resulting phase data has a higher signal-to-noise ratio and a more uniform stripe density.

[0059] Step 10: Pixel-level matching

[0060] Using the expanded phase values, a one-to-one mapping relationship between camera pixel coordinates and screen pixel coordinates is established (usually an intermediate phase plane is introduced to achieve fast lookup).

[0061] Step 11: Ray tracing to reconstruct the 3D topography

[0062] Based on the calibrated camera-screen geometric relationship and the pixel correspondence in step 10, the three-dimensional coordinates of each point on the measured mirror are calculated by reverse ray tracing, and finally the complete three-dimensional shape is obtained.

[0063] In one specific embodiment:

[0064] First, the parameters of the two cameras and the projection screen in the system are calibrated using the flat panel calibration method, and the transformation relationships between the camera 1 coordinate system, camera 2 coordinate system, screen coordinate system and world coordinate system are established.

[0065] Then, project a set of periods onto the display screen. Horizontal and vertical long-period equally spaced stripes with different initial phases were captured by two cameras after reflection from the SUT (Surface Mount Technology). A multi-step phase-shifting algorithm and a spatial phase unwrapping algorithm were used to initially establish the correspondence between camera pixels and display pixels. Simultaneously, based on the phase gradient distribution of the stripe images captured by the cameras in the horizontal and vertical directions, the phase gradient distribution of the stripes on the display was set, thus forming adaptive stripes in both the horizontal and vertical directions on the display, which were then reprojected.

[0066] Using the same solution method, a precise correspondence between each camera pixel and the display pixel under adaptive fringes is established, and then the high-precision three-dimensional shape of the SUT can be reconstructed by ray tracing.

[0067] This application embodiment further provides the solution process for adaptive stripes and the solution process for the precise pixel correspondence between the camera and the screen, using camera 1 as an example for illustration.

[0068] First, when the display shows horizontal and vertical large-period equally spaced stripes, the camera 1 captures the corresponding image and solves the phase. and Thus, the camera pixels are obtained. Gradients in the horizontal and vertical directions:

[0069] (1)

[0070] Secondly, according to and Values ​​and adaptive threshold T G The stripe period of each pixel in the display plane is either sparsified or densified. That is, after mapping each pixel in the camera plane back to the display plane, its adaptive stripe gradient distribution in the display plane is calculated according to formula (2). Finally, through methods such as interpolation, the adaptive stripe gradient distribution of the entire display plane is established. and .

[0071] The horizontal and vertical phase distributions are obtained by integrating the phase gradient of the display plane. and Then, sinusoidal modulation can be performed to generate an adaptive fringe pattern.

[0072]

[0073] in, The pixel coordinates of the display plane.

[0074] Then, by re-measuring using adaptive fringes, the updated horizontal and vertical phase distributions can be obtained. and .

[0075] Furthermore, to accurately obtain the correspondence between pixels on the camera plane and the display plane, an intermediate phase plane is introduced, such as... Figure 5 As shown.

[0076] First, establish a phase plane, where each phase point... The corresponding set of horizontal stripe phases Phase with vertical stripes The transformation relationship is shown in formula (3):

[0077]

[0078] in, This represents the phase value per unit pixel within the phase plane.

[0079] Secondly, all pixels in the display plane are mapped to the phase plane according to their horizontal and vertical phases, and a lookup table from the phase plane to the display plane is established using the difference and search methods. Their relationship can be expressed by formula (4):

[0080] (4)

[0081] Finally, based on the camera plane pixels The phase is obtained, and its corresponding coordinate point in the phase plane is obtained. Then, based on the lookup table relationship between the phase plane coordinates and the display plane pixels... This allows us to obtain the corresponding display plane pixel. Their relationship can be expressed by formula (5):

[0082]

[0083] In summary, this application proposes an adaptive fringe resolution-based PMD method by modifying the projected fringe pattern. This method, without adding any measurement components, adapts the local density of the fringe displayed on the projection screen to the angular transformation characteristics of the reflected light from the measured surface. This ensures that the distribution of the structured light fringes measured from the camera's perspective is uniform and appropriately sparse. Furthermore, while maintaining good analytical integrity, it maximizes the dynamic range and accuracy of the PMD under the same measurement conditions by improving phase resolution accuracy. This effectively solves the problems of low measurement accuracy and local surface shape measurement failure caused by excessive curvature of the local measurement surface in conventional PMDs. Moreover, this method can be combined with other existing PMDs to simultaneously optimize both the measurement components and the projected fringe, expanding its measurement capabilities.

[0084] It should be noted that the above content merely illustrates the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. For those skilled in the art, various improvements and modifications can be made without departing from the principle of the present invention, and all such improvements and modifications fall within the scope of protection of the claims of the present invention.

Claims

1. A PMD three-dimensional measurement method with adaptive fringe projection mode, characterized in that, include: a) Based on the surface morphology features of the mirror element under test, obtain its phase gradient distribution under the camera's view. b) Based on the phase gradient distribution, the local period of the projected stripes is adaptively modulated so that the structured light stripes formed on the camera imaging surface after reflection from the measured surface remain uniform and moderately sparse. c) Project an adaptive fringe pattern and acquire a fringe image modulated by the surface under test to establish the correspondence between camera pixels and projection pixels; d) Based on the aforementioned correspondence, the three-dimensional morphology of the tested mirror element is reconstructed by ray tracing.

2. The method according to claim 1, characterized in that, The phase gradient distribution is obtained by first projecting a set of equal-period horizontal and vertical large-period stripes and then analyzing the images captured by the camera.

3. The method according to claim 1 or 2, characterized in that, The adaptive modulation step includes: mapping the phase gradient of the camera plane back to the display plane, and sparsening or densifying the stripe period of each region in the display plane using a preset adaptive threshold.

4. The method according to claim 3, characterized in that, The process of mapping the phase gradient of the camera plane back to the display plane specifically involves using the calibrated camera-screen geometric relationship to reverse-map the coordinates of each pixel on the camera imaging plane to the pixel coordinates on the display plane.

5. The method according to claim 3, characterized in that, Specifically, the stripe period of each region within the display plane is thinned or densified using a preset adaptive threshold. A. Determine the density of local fringes: Set a threshold for the phase gradient and compare the gradient value at each location with this threshold. If the gradient value is lower than the threshold, it means that the stripes in the corresponding region are too sparse in the camera's view. If the gradient value is higher than the threshold, it means that the stripes are too dense; B. Calculate the new local period: In overly sparse regions, the fringe period is reduced; in overly dense regions, the fringe period is increased. The new local period is inversely proportional to the measured phase gradient; the larger the gradient, the larger the period; the smaller the gradient, the smaller the period, thus ensuring that the fringe density in all regions of the camera tends to be consistent. C. Generate a new phase distribution: Convert the adjusted period into a phase gradient field in the display plane, and then integrate along the horizontal and vertical directions to obtain the phase value corresponding to each pixel of the display. This phase distribution is the phase skeleton of the adaptive stripe.

6. The method according to claim 1 or 2, characterized in that, The projection using an adaptive fringe pattern specifically involves: modulating the phase skeleton into a sinusoidal grayscale fringe pattern, generating one in the horizontal and one in the vertical directions, and projecting them sequentially onto the mirror surface under test.

7. The method according to claim 6, characterized in that, The establishment of the correspondence between camera pixels and projection pixels introduces an intermediate phase plane, and a precise mapping is achieved through a lookup table between the phase plane and the display plane.

8. The method according to claim 1, characterized in that, The ray tracing is based on a dual-camera stereo vision model, and the three-dimensional coordinates of each point on the measured surface are calculated by combining camera-screen calibration parameters.

9. The method according to claim 1, characterized in that, The mirror element under test is an optical surface with large local curvature or a complex freeform surface.

10. The method according to claim 1, characterized in that, The adaptive fringe pattern can be dynamically refreshed during a single measurement to iteratively optimize the fringe density and further improve the phase resolution accuracy.

Images