CN-121977477-A - Stripe projection high-reflection surface phase unwrapping method based on frequency multiplexing
Abstract
The invention relates to a fringe projection high-reflection surface phase unwrapping method based on frequency multiplexing, which comprises the steps of collecting a multi-frequency fringe image, dividing the multi-frequency fringe image into overexposed areas and non-overexposed areas, obtaining wrapping phases in the non-overexposed areas by adopting a generalized phase shift method, obtaining initial unwrapped phases based on a multi-frequency heterodyne algorithm, obtaining continuous unwrapped phases containing the high-reflection areas through block filtering and interpolation processing, constructing complementary gray codes consistent with wrapping phase periods by combining multi-threshold constraint, correcting fringe orders of the overexposed areas, further carrying out brightness attenuation design on medium-low frequency fringes, realizing frequency multiplexing, obtaining unwrapped phases of overexposed points in the non-high frequency fringes, mapping the unwrapped phases to high frequencies, extracting and fusing the wrapping phases by adopting one-dimensional Fourier and Hilbert transformation, and selecting and inhibiting spectrum leakage through a complete period, thereby realizing effective correction on the wrapping phases and the fringe orders of the overexposed points and obtaining reliable unwrapped phases.
Inventors
- LI WENJIE
- HUANG YUYUAN
- WANG HAIJIAN
- LIU GUIJIE
- XU CHENYANG
Assignees
- 桂林电子科技大学
Dates
- Publication Date
- 20260505
- Application Date
- 20260123
Claims (10)
- 1. The fringe projection high-reflection surface phase unwrapping method based on frequency multiplexing is characterized by comprising the following steps of: acquiring a multi-frequency phase shift fringe image of an object to be detected, carrying out pixel-level intensity analysis on the highest frequency fringe image in the multi-frequency phase shift fringe image, detecting and marking an overexposure region with saturated intensity, dividing the image into an overexposed region and a non-overexposed region, and solving a wrapping phase of the non-overexposed region by adopting a generalized phase shift method for obtaining a medium-frequency and low-frequency wrapping phase and an overexposed region mask; According to the multi-frequency heterodyne principle, an initial unfolding phase of each frequency is obtained, a reliable unfolding phase point is obtained through modulation mask filtering, plane blocking filtering and median filtering operation, and then a continuous unfolding phase containing a high reflection area is respectively obtained at each frequency through a data interpolation algorithm; Performing multi-threshold constraint on the complete continuous unfolding phase under each frequency, constructing complementary Gray codes consistent with the wrapping phase period, and correcting the fringe order of each frequency unfolding phase; When the projection pattern is designed, the intermediate frequency projection fringe pattern and the low frequency projection fringe pattern are multiplied by a preset coefficient smaller than 1, namely, the target darkness is set for the non-high frequency pattern under the same condition, so that the point which is expected to be overexposed at the high frequency is not overexposed any more, then the unfolding phase of the point is obtained at the non-high frequency, and then the unfolding phase is converted to the high frequency unfolding phase according to the frequency relation, so that the frequency multiplexing is realized; And for points with all frequencies overexposed, extracting fundamental frequency in a certain area of a line where the points are located by adopting one-dimensional Fourier transform, then obtaining a new unexposed fringe pattern by using inverse Fourier transform, carrying out wrapping phase calculation, carrying out Hilbert transform on the obtained one-dimensional fringe pattern, then carrying out wrapping phase calculation, and fusing the two wrapping phase calculation for carrying out high-precision correction on the wrapping phase of the overexposed points.
- 2. The frequency reuse based fringe projection high reflection surface phase unwrapping method of claim 1, wherein obtaining said overexposed area mask comprises: And (3) carrying out pixel-by-pixel analysis on the multi-frequency phase shift fringe image, comparing the gray value of each frame of a pixel point with an overexposure threshold value aiming at the fringe image with a certain frequency, defining that the pixel point is overexposed in the frame when the gray value exceeds the threshold value, marking the pixel point as an overexposed point when the non-overexposed frame number of the pixel point is less than 3, processing other pixel points in the same way, and then respectively establishing an overexposed mask and a non-overexposed mask at each frequency.
- 3. The fringe projection high reflection surface phase unwrapping method based on frequency multiplexing as recited in claim 1, comprising, after obtaining said continuous unwrapped phase: The method comprises the steps of dividing an entire initial unfolded phase region into four regions based on slope characteristics of unfolded phases, carrying out plane filtering based on random consistency, setting a point threshold value in a plane to be 5 and less than a fringe order phase range 2 pi so as to remove fringe order error points, and obtaining accurate phase fringe orders when complementary gray code construction is carried out.
- 4. The frequency-multiplexed based fringe projection high-reflectivity surface phase unwrapping method of claim 1, wherein constructing the complementary gray codes comprises: Normalizing the expansion phase obtained by interpolation to a [ -pi, pi ] interval based on the expansion phase range of the projection pattern, and then carrying out the following multi-threshold constraint on the initial expansion phase of each frequency to construct a complementary gray code: ; ; Wherein, the For the gray code pattern sequence, To normalize to the unwrapped phase in the [ -pi, pi ] interval, For the index number of the index, Is the m-th frame gray code pattern.
- 5. The frequency-multiplexed based fringe projection high reflection surface phase unwrapping method of claim 1 wherein correcting the fringe order of each frequency unwrapped phase comprises: and correcting the stripe grade of the overexposed point in the overexposed region, namely obtaining the correct stripe grade by utilizing a complementary Gray code decoding algorithm according to the constructed Gray code pattern, and correcting the stripe grade of the overexposed point in the overexposed region.
- 6. The fringe projection high reflection surface phase unwrapping method based on frequency multiplexing of claim 1, wherein implementing said frequency multiplexing comprises: For the pixel points which are overexposed at high frequency and not overexposed at medium frequency or low frequency, the proportional relation among the frequencies is utilized, and based on the unfolding phase of the point at the medium frequency or the low frequency, the unfolding phase of the point at the high frequency is recovered, so that the frequency multiplexing is realized: for the pixel points which are overexposed at high frequency but not overexposed at medium frequency, the high-frequency unfolding phase is corrected as follows: ; Wherein, the Is a medium frequency (intermediate frequency), In order to be at a high frequency, For the unwrapped phase of this point after the high frequency correction, Developing a phase for the intermediate frequency of the point; for pixels that are overexposed at high and medium frequencies, but not overexposed at low frequencies, the high frequency unwrapped phase is modified as: ; Wherein, the Is a frequency of a low frequency and, For the unwrapped phase of this point after the high frequency correction, The phase is spread out for this point at low frequency.
- 7. The frequency-multiplexed based fringe projection high reflection surface phase unwrapping method of claim 1, wherein calculating said wrapped phase comprises: For pixel points with all frequencies overexposed, a new fringe diagram is constructed based on one-dimensional Fourier transform and one-dimensional Hilbert transform, and wrapping phases respectively obtained by the two fringe diagrams are fused, so that the wrapping phase correction of the overexposed points is realized: Stripe projection is vertical stripe projection, and for pixel points (x 0, y 0) with all frequencies being overexposed, local one-dimensional signals are extracted along the row where the pixel points are located: ; Wherein, the For the one-dimensional stripe signal corresponding to the n-th phase shift extracted in the saturated pixel neighborhood, L is half window length of the local one-dimensional stripe signal and is used for limiting the one-dimensional stripe extraction range, and the value of L is characterized in that the period size is obtained in a non-overexposure area, and L is an integer multiple of the period size so as to prevent spectrum leakage during one-dimensional Fourier transform.
- 8. The frequency-multiplexed based fringe projection high reflection surface phase unwrapping method of claim 1, wherein extracting said fundamental frequency comprises: through carrying out Fourier transformation on the local one-dimensional stripe signals, a windowing filter is constructed based on fundamental frequency positioning, direct current components and higher harmonics are restrained, and then fundamental frequency components in the signals are extracted: ; ; Wherein: ; In the formula, For the one-dimensional stripe signal corresponding to the n-th phase shift extracted in the saturated pixel neighborhood, Is a one-dimensional stripe signal Is provided with a spectral representation of (a), For the windowed filtered spectrum, Is a frequency domain window function, is used for restraining direct current components and higher harmonic components, Is a frequency variable in a one-dimensional fourier transform, Is the corresponding fundamental frequency position of the one-dimensional stripe signal, Half-bandwidth parameters that are frequency domain window functions; the inverse transformation results in a filtered fringe, and then a fourier filtered phase is calculated: ; ; Wherein, the For the wrapped phase calculated based on the one-dimensional fourier filter fringe pattern, For the windowed filtered spectrum, Is a one-dimensional inverse Fourier transform operator, and N is the phase shift step number.
- 9. The frequency-multiplexed based fringe projection high reflection surface phase unwrapping method of claim 7 wherein said one-dimensional hilbert transform is used to construct an resolved signal orthogonal to the filtered one-dimensional fringe signal to obtain a wrapped phase complementary to the fourier filter phase error, comprising: ; ; Wherein, the For the wrapped phase calculated based on the one-dimensional hilbert transformed fringes, Is a one-dimensional Hilbert transform operator, and N is the phase shift step number.
- 10. The frequency-multiplexing-based fringe projection high-reflectance surface phase unwrapping method of claim 1, wherein high-precision correcting the overexposed point wrapping phase comprises: the local phase fusion is realized by averaging two groups of wrapping phases: ; Wherein, the To obtain a corrected wrapped phase by local phase fusion, For the wrapped phase calculated based on one-dimensional fourier filter fringes, The wrapping phase is calculated based on one-dimensional Hilbert transformation stripes; And extracting the wrapping phase of the overexposure point from the fusion phase, and combining the strip gradation correction of the overexposure point to realize the unfolding phase correction of the overexposure point.
Description
Stripe projection high-reflection surface phase unwrapping method based on frequency multiplexing Technical Field The invention relates to the technical field of optical three-dimensional measurement, in particular to a fringe projection high-reflection surface phase unwrapping method based on frequency multiplexing. Background Stripe projection profile (FPP) is widely applied to the fields of industrial detection and quality control, reverse engineering, biomedical engineering, cultural heritage preservation, digitization and the like due to the advantages of high precision, high efficiency, non-contact and the like. The mainstream FPP technology can be generally divided into two types, fourier Transform Profilometry (FTP) and Phase Shift Profilometry (PSP), and has been widely used in practical applications because the PSP technology has stronger robustness to phase noise caused by ambient light and surface reflectivity, and can implement pixel-by-pixel phase measurement results with higher resolution and accuracy. However, when a high reflectivity region exists on the surface of the measurement object or the illumination condition of the measurement environment is complex, the stripe image collected by the camera is easy to generate a light intensity saturation phenomenon due to the limited dynamic range of the camera. In the saturated region, the gray value of the image exceeds the quantization range of the camera, so that partial information is lost, and the phase value calculated based on the conventional phase shift algorithm can generate serious errors, thereby affecting the accuracy and the integrity of three-dimensional reconstruction. Aiming at the problems, the prior solutions are mainly divided into two types, namely a method based on hardware equipment adjustment and a technology based on algorithm. The former mainly includes multiple exposure techniques and adaptive projection methods. Such methods are essentially inverse multiple exposure techniques, requiring pre-projection of a specific pattern to perceive the reflective properties of the object surface. However, the number of projection patterns required is generally large, resulting in a decrease in measurement efficiency. The latter focuses on directly calculating the object phase using the saturation fringes and correcting the errors introduced by the fringe saturation, increasing the number of phase steps or using the inverse fringes can extend the dynamic range of the system to some extent. However, too small a number of stripes may limit the dynamic range of the dynamic range, while too large a number of stripes may increase the data processing time. Therefore, a phase unwrapping method is needed that can use multi-frequency information complementation based on the frequency multiplexing principle in a single set of exposure data, and eliminate overexposure interference through stripe order correction and wrapping phase correction of overexposure points, so as to improve measurement accuracy and robustness of a highly reflective object. Disclosure of Invention Aiming at the problems of multiple projection patterns, complicated operation, poor robustness and the like in the phase unwrapping of a high-reflection object in the prior art, the invention provides a stripe projection high-reflection surface phase unwrapping method based on frequency multiplexing, which corrects stripe orders and wrapping phases of overexposure points through multi-frequency stripe information fusion, realizes accurate phase recovery of a high-reflection area, does not need additional hardware support, and is compatible with the existing stripe projection system. In order to achieve the above object, the present invention provides the following solutions: A fringe projection high-reflection surface phase unwrapping method based on frequency multiplexing comprises the following steps: acquiring a multi-frequency phase shift fringe image of an object to be detected, carrying out pixel-level intensity analysis on the highest frequency fringe image in the multi-frequency phase shift fringe image, detecting and marking an overexposure region with saturated intensity, dividing the image into an overexposed region and a non-overexposed region, and solving a wrapping phase of the non-overexposed region by adopting a generalized phase shift method for obtaining a medium-frequency and low-frequency wrapping phase and an overexposed region mask; According to the multi-frequency heterodyne principle, an initial unfolding phase of each frequency is obtained, a reliable unfolding phase point is obtained through modulation mask filtering, plane blocking filtering and median filtering operation, and then a continuous unfolding phase containing a high reflection area is respectively obtained at each frequency through a data interpolation algorithm; Performing multi-threshold constraint on the complete continuous unfolding phase under each frequency, constructing complementa