CN115861542A - Binocular multiline three-dimensional reconstruction method and system - Google Patents
Binocular multiline three-dimensional reconstruction method and system Download PDFInfo
- Publication number
- CN115861542A CN115861542A CN202211621678.2A CN202211621678A CN115861542A CN 115861542 A CN115861542 A CN 115861542A CN 202211621678 A CN202211621678 A CN 202211621678A CN 115861542 A CN115861542 A CN 115861542A
- Authority
- CN
- China
- Prior art keywords
- line
- laser
- modulated
- image
- structured light
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention relates to a binocular multiline three-dimensional reconstruction method and a binocular multiline three-dimensional reconstruction system, wherein the method comprises the following steps: projecting speckles to a measured object by using a laser speckle projector, and collecting a speckle image pair modulated on the surface of the measured object; performing epipolar line correction on the modulated speckle image pair and calculating matching cost to obtain a disparity map; performing multi-line structured light scanning on a measured object through a single-line laser and a rotary galvanometer, and sequentially acquiring a plurality of modulated multi-line structured light images on the surface of the measured object in the scanning process; performing epipolar line correction on a plurality of modulated multi-line structured light images, and performing sub-pixel central point extraction on light bars; carrying out light bar matching according to the parallax map, and determining the light bar matching relation in the multiple modulated multi-line structured light images; and generating point cloud data by utilizing a triangular distance measurement principle according to the sub-pixel coordinates of each central point in the light bar, and completing the three-dimensional reconstruction of the measured object.
Description
Technical Field
The invention relates to the field of machine vision, in particular to a binocular multiline three-dimensional reconstruction method and system.
Background
The three-dimensional reconstruction refers to the restoration and reconstruction of some three-dimensional objects or three-dimensional scenes, and the reconstructed model is convenient for computer representation and processing. In the related art, a structured light method, a Time of flight (TOF) method, a Multiple View Stereo (MVS) method, and the like may be employed. The structured light three-dimensional reconstruction technology calculates information such as the position and depth of an object according to the change of an optical signal caused by the object, has the advantages of high measurement precision, non-contact, short imaging time and the like, and is widely applied to the fields of industrial detection, reverse engineering, virtual reality, dental diagnosis, cultural relic protection and the like.
In the structured light three-dimensional reconstruction technique, a surface structured light method, a speckle structured light method, a single line laser structured light method, or the like can be employed. When the surface projection structured light method is adopted, the surface projection structured light must be projected by using a projection light machine, and the projection light machine has the defects of large volume, large heat productivity, high price and the like. The speckle structured light method realizes three-dimensional reconstruction through binocular matching of laser speckles, has large calculated amount and low reconstruction precision, and cannot meet the application requirement of high-precision industrial detection. The single line laser structured light method projects line structured light to the surface of a measured object through a laser, the line structured light is reflected to a camera through a modulated light bar on the surface of the object to obtain a modulated light bar image, only one light bar can be reconstructed from one image, therefore, line structured light images at different positions of the measured object need to be collected, the three-dimensional reconstruction can be completed only by collecting multiple light bar images through the camera, and the single line laser structured light method has the defects of low working efficiency, high transmission bandwidth requirement and the like.
The binocular three-dimensional reconstruction method based on the multi-line structured light has the advantages of small volume, low cost, high efficiency, high precision, low bandwidth and the like, and can solve the problems of low industrial adaptability of the multi-line structured light and mismatching of light bars. The chinese patent documents CN111854642B and CN114723828A both adopt a scheme that a multi-line laser with a multi-line DOE module is matched with a rotating reflector, and when the DOE module is used to realize multi-line laser projection, the central bright spot problem may exist, which affects the extraction precision of the subsequent laser line central point; the DOE module is used for realizing that laser lines projected by the multi-line laser are too small and cannot be changed, the problem of repeated scanning of partial regions exists, and the advantage of high efficiency of the multi-line laser cannot be fully exerted. Patent publication CN114782632A adopts a linear structured light back projection combining with a feature point texture fusion matching method, and needs to calibrate linear structured light parameters, which is complicated to calculate.
Disclosure of Invention
The invention aims to provide a binocular multi-line three-dimensional reconstruction method and a binocular multi-line three-dimensional reconstruction system, which are used for solving the problems that the speckle structured light method is large in calculated amount and low in reconstruction precision and cannot meet the application requirement of high-precision industrial detection; the single-line laser structure optical method has the problems of low working efficiency and high transmission bandwidth requirement; the problem of central bright spots can exist in the multi-line laser projection realized by the DOE module, the extraction precision of the central point of the follow-up laser line is influenced, the multi-line laser projection laser line is too small and cannot be changed by the DOE module, repeated scanning of partial areas exists, and the advantage of high efficiency of the multi-line laser cannot be fully played.
In order to solve the problems, the invention adopts the following technical scheme:
a binocular multiline three-dimensional reconstruction method comprises the following steps:
s1: projecting speckles to a measured object by using a laser speckle projector, and collecting a speckle image pair modulated on the surface of the measured object;
s2: performing epipolar line correction on the modulated speckle image pair and calculating matching cost to obtain a disparity map;
s3: performing multi-line structured light scanning on the object to be measured through a single-line laser and a rotary galvanometer, and sequentially acquiring a plurality of modulated multi-line structured light images on the surface of the object to be measured in the scanning process;
s4: performing epipolar line correction on a plurality of modulated multi-line structured light images, and performing sub-pixel central point extraction on light bars;
s5: carrying out light bar matching according to the parallax map, and determining the light bar matching relation in the multiple modulated multi-line structured light images;
s6: and generating point cloud data by utilizing a triangulation principle according to the sub-pixel coordinates of each central point in the light bar, and finishing the three-dimensional reconstruction of the measured object.
Further, step S2 specifically includes: performing epipolar line correction on the modulated speckle image pair to obtain a corrected left image and a corrected right image; creating an M-by-M region by taking an optional point in the left image as a center, searching the region with the same size in the right image along an epipolar line, and calculating the matching cost once, wherein the cost calculation formula is as follows:
C(u,v,d)＝ρ(C AD (u,v,d),λ AD )+ρ(C ce (u,v,d),λ ce )
wherein, C AD (u, v, d) is the matching cost obtained by the AD method; c ce (u, v, d) is the hamming distance based on Census transform;
is a normalized formula, where c is the matching cost value and λ is the control parameter; and obtaining the disparity map according to a cost calculation formula.
Further, step S4 specifically includes: performing epipolar line correction on any modulated multi-line structured light image, obtaining an optical strip edge image by applying a Canny operator, performing expansion processing on the edge image and obtaining an outermost layer profile; and calculating the center point of the sub-pixel in the area in the contour by using a gray scale gravity center method, and smoothing the spline of the center point of the sub-pixel of the light bar by cubic spline interpolation.
The invention also provides a binocular multiline three-dimensional reconstruction system, which comprises a left camera, a right camera, a single-line laser, a laser speckle projector, a rotary galvanometer and a control unit, wherein the control unit respectively controls the left camera, the right camera, the single-line laser, the laser speckle projector and the rotary galvanometer, the control unit comprises a processor and a memory, and the processor processes a program stored in the memory to realize the following steps:
s1: projecting speckles to a measured object by using a laser speckle projector, and collecting a speckle image pair modulated on the surface of the measured object;
s2: performing epipolar line correction on the modulated speckle image pair and calculating matching cost to obtain a disparity map;
s3: performing multi-line structured light scanning on the object to be detected through a single-line laser and a rotary galvanometer, and sequentially acquiring a plurality of modulated multi-line structured light images on the surface of the object to be detected in the scanning process;
s4: performing epipolar line correction on a plurality of modulated multi-line structured light images, and performing sub-pixel central point extraction on light bars;
s5: carrying out light bar matching according to the disparity map, and determining the light bar matching relation in the multiple modulated multi-line structured light images;
s6: and generating point cloud data by utilizing a triangulation principle according to the sub-pixel coordinates of each central point in the light bar, and finishing the three-dimensional reconstruction of the measured object.
Further, step S1 specifically includes: the control unit controls the rotary galvanometer to move to a zero position, and when the rotary galvanometer moves to the zero position, the control unit controls to start the laser speckle projector, sends collected image pulse signals to the left camera and the right camera and obtains a speckle image pair modulated on the surface of the measured object.
Further, step S2 specifically includes: performing epipolar line correction on the modulated speckle image pair to obtain a corrected left image and a corrected right image; creating an M-by-M region by taking an optional point in the left image as a center, searching the region with the same size in the right image along an epipolar line, and calculating the matching cost once, wherein the cost calculation formula is as follows:
C(u,v,d)＝ρ(C AD (u,v,d),λ AD )+ρ(C ce (u,v,d),λ ce )
wherein, C AD (u, v, d) is the matching cost obtained by the AD method; c ce (u, v, d) is the hamming distance based on Census transform;
is a normalized formula, where c is the matching cost value and λ is the control parameter; obtaining the vision according to a cost calculation formulaAnd (4) a difference graph.
Further, step S3 specifically includes: the control unit controls the single-line laser and the rotary galvanometer to perform multi-line structured light scanning on the measured object, simultaneously sends acquired image pulse signals to the left camera and the right camera, and controls the rotary galvanometer to rotate to open the single-line laser within the exposure time of the cameras and close the single-line laser after t ms; then the control unit controls the rotating position L + S of the rotary galvanometer, then the single-line laser is turned on, and the single-line laser is turned off after t ms; then the control unit controls the rotating position L +2S of the rotary galvanometer, then the single-line laser is turned on, and is turned off after t ms; and in the same way, obtaining a plurality of modulated multi-line structured light images, wherein L is the position of a first light strip projected, and S is the line spacing or the laser line width.
Further, step S4 specifically includes: performing epipolar line correction on any modulated multi-line structured light image, obtaining an optical strip edge image by applying a Canny operator, performing expansion processing on the edge image and obtaining an outermost layer profile; and calculating the center point of the sub-pixel in the area in the contour by using a gray scale gravity center method, and smoothing the spline of the center point of the sub-pixel of the light bar by cubic spline interpolation.
The invention has the beneficial effects that:
the invention projects single line laser through the single line laser, projects the single line laser to the surface of the measured object through the rotating prism, controls the prism to rotate and the laser to be switched on and off within the simultaneous exposure time of the left camera and the right camera, and finishes multi-line scanning of the measured object. The method for matching the parallax map with the optical strips is high in accuracy, and optical parameters of the multi-line structure do not need to be calibrated. In the traditional scheme, the left view point is re-projected to the right view through the multi-line structured light parameter, the similarity of the point pair to be matched is calculated, and the matched point pair is determined; the disparity map is used as an index, and the point matching is replaced by the optical strip matching, so that the matching precision can be effectively improved, and the structural optical parameters do not need to be calibrated.
Drawings
FIG. 1 is a schematic structural diagram of a binocular multiline three-dimensional reconstruction system of the present invention;
FIG. 2 is a flow chart of a binocular multiline three-dimensional reconstruction method of the present invention;
FIG. 3 is an optical representation of a laser speckle structure of the present invention;
FIG. 4 is a schematic diagram of the control process of the present invention;
FIG. 5 is a schematic diagram of a single line laser and a rotary galvanometer control process of the present invention;
fig. 6 is a schematic diagram of the principle of matching point-to-triangle distance measurement according to the present invention.
Detailed Description
Binocular multiline three-dimensional reconstruction system embodiment:
as shown in fig. 1 to 6, the binocular multi-line three-dimensional reconstruction method and system provided by the present invention includes a left camera, a right camera, a single-line laser, a laser speckle projector, a rotary galvanometer, and a control unit, where the control unit includes a data processing unit, and the control unit controls the left camera, the right camera, the single-line laser, the laser speckle projector, and the rotary galvanometer, respectively.
When the system works, the first step is as follows: as shown in fig. 4, the control unit controls the rotary galvanometer to move to a zero position, when the galvanometer moves to the zero position, the control unit starts the laser speckle projector, sends pulse signals for collecting images to the left camera and the right camera, and obtains a speckle image pair I modulated by a measured object l0 、I r0 The control unit turns off the laser speckle projector. Fig. 3 is an optical schematic of a laser speckle structure of the present invention.
The second step is: according to the left and right camera internal parameters E and the left and right camera position relation R and T, the modulated speckle image I is processed l0 、I r0 Polar line correction to obtain polar line corrected image IR l0 、IR r0 And the epipolar lines of the two images are exactly in the same horizontal line, and the corresponding feature points only need to be searched in the same line.
To correct the left image IR l0 Creating a region of M (M is an odd number) centered at an optional point, and correcting the right image IR along the epipolar search r0 Sequentially calculating and matching the region with the same size M x MAnd (3) determining a matching point according to the matching cost, wherein the cost calculation is realized by combining an AD method and a Census method.
The AD method obtains the matching cost by calculating the sum of absolute interpolation of gray scales of the to-be-matched areas of the left image and the right image, and the calculation formula is
The Census method is to obtain the matching cost by transforming Census of the left and right images to be matched and calculating the hamming distance. Census conversion is to convert pixel gray into a bit string using local gray differences in the neighborhood of pixels, to map a boolean value obtained by comparing the gray value of a pixel in the neighborhood (region size M × M) with the gray value of the center pixel into one bit string, and to finally use the value of the bit string as the Census conversion value of the center pixel, the Census conversion formula being as follows
Wherein m' is the largest integer no greater than half m,
for bitwise concatenation of bits, the ξ operation is defined as follows:
the Hamming distance is the number of the two bit strings with different corresponding bits, and the calculation method is to perform OR operation on the two bit strings and count the number of the bit bits of the OR operation result which is not 1.
C ce (u,v,d)＝Hamming(C sl (u,v),C sl (u+d,v))
The AD method and the Census method are combined together through normalization, and the normalization formula is as follows:
c is the matching cost value, lambda is the control parameter, when c, lambda is positive value, the function value interval is [0,1].
Finally, the cost is calculated by the formula
C(u,v,d)＝ρ(C AD (u,v,d),λ AD )+ρ(C ce (u,v,d),λ ce )
Calculating a corrected image IR according to a cost calculation formula l0 、IR r0 After that, the corrected image IR is l0 、IR r0 Calculating a right parallax map by position exchange; calculating a pixel position b of a homonymy point in the right disparity map for each pixel a of the left disparity map; and judging whether the absolute value of the difference between the parallax values of a and b is less than a threshold (usually 1 pixel), if so, failing to pass the consistency check, and changing the parallax of the corresponding position into an invalid value.
Filling the invalid value position in the left disparity map with the minimum value in the field, and finally applying median filtering to the left disparity map to obtain the optimized left disparity map ID.
And step three, as shown in fig. 5, controlling the single-line laser and the rotary galvanometer to perform multi-line structured light scanning on the object to be measured. The control unit simultaneously sends collected image pulse signals to the left camera and the right camera, and in the exposure time of the cameras, the control unit controls the rotary position L of the rotary galvanometer to turn on the laser and turns off the laser after t ms; the control unit controls the rotating position L + S of the rotary galvanometer and turns on the laser, and turns off the laser after t ms; the control unit controls the rotating position L +2S of the rotary galvanometer and turns on the laser, and turns off the laser after t ms; and so on. The single-line laser is reflected by the rotary vibrating mirrors at different positions and then irradiates the surface of the measured object, the reflected light on the surface of the object at different moments is received by the camera, and finally a second modulated multi-line structured light image I is obtained l1 、I r1 Wherein L is the position of the first light bar projected and S is the line spacing.
Then, the control unit sends pulse signals for collecting images to the left camera and the right camera simultaneously,in the exposure time of the camera, the control unit controls the rotating position L + S of the galvanometer to turn on the laser, and turns off the laser after t ms; the control unit controls the rotation position L + S + W of the galvanometer and turns on the laser, and turns off the laser after t ms; the control unit controls the rotation position L +2S + W of the galvanometer and turns on the laser, and the laser is turned off after t ms; and so on. The single-line laser is reflected by the vibrating mirrors at different positions and then irradiates the surface of the measured object, the reflected light on the surface of the object at different moments is received by the camera, and finally a second modulated multi-line structured light image I is obtained l2 、I r2 Wherein W is the laser linewidth.
And so on until the control unit controls the camera to acquire an S/W multi-line modulation image I lS/W 、I rS/W And at this point, completing the image acquisition process.
The fourth step is: correcting the polar line of any modulation multi-line image, and solving an optical strip edge image by applying a Canny operator; the edge image is dilated and the outermost contour is acquired. Calculating the center point of the sub-pixel of the light bar by using a gray scale gravity center method for the area in the outline; the sub-pixel center point spline for each bar is smoothed by cubic spline interpolation.
The fifth step is: based on disparity map ID and modulation multiline left image I li A sub-pixel central point set P of a certain light bar li Calculating a modulated multiline right image I ri Corresponding point set P' li . Calculating modulated multiline right image I ri Sub-pixel center point set P of middle light bar r1 、P r2 、P r3 …P rj …P rn And form a right light stripe set P r . When positioning point set P' li A point in the middle and a set of right light bar points P rj The distance between the polar lines of a certain point is less than a threshold value V d (typically two pixels), then the two points are considered to match; when positioning point set P' li Number of point matches V between midpoint and a set of right light bar points m Maximum and V m Greater than positioning point set P' li And if the total point number is one half, the two point sets are considered to be matched, namely the two light bars are matched. According to the method, an image I is determined l1 、I r1 、I l2 、I r2 ……I lN/P 、I rN/P Middle line pieceAnd (4) matching relationship.
The sixth step is: as shown in FIG. 6, which is a schematic diagram of the triangularization method, the position of a matching point pair on the image plane of the left camera and the right camera is P l 、P r And the coordinate of the three-dimensional point P is the corresponding three-dimensional point of the matching point pair. Obtaining a P coordinate by a triangular distance measurement principle:
wherein, parallax D = X l -X r (ii) a f, the focal length of the camera can be obtained from the internal reference of the calibrated camera; and B is the base line distance of the left camera and the right camera, and can be obtained from the calibrated cameras.
And traversing all the matching point pairs to obtain three-dimensional point cloud data so as to complete three-dimensional reconstruction.
For example, the left camera and the right camera can adopt black-and-white or color cameras, and filters for covering the wavelengths of the laser speckle projector and the single-line laser are added in the cameras, so that only light in a required structural light wavelength range is allowed to pass, namely, only the wavelength of the structural light reflected by the object to be measured is received, a reflected image of the structural light is obtained, the contrast is improved, and the interference of ambient light is reduced.
Illustratively, the single-line laser is a laser capable of emitting a uniform laser line, and may be composed of a laser diode and a powell prism.
Illustratively, a laser speckle projector is used to project speckle structured light onto the target object under test, the speckle wavelength can be 450nm, 520nm, 658nm, or other wavelengths, see fig. 2.
Illustratively, the rotating galvanometer causes the laser line to scan in translation on the surface of the object.
Illustratively, the data processing unit may be a processor, such as a CPU or GPU, or the like, and the data processing unit may process the acquired image, such as the modulated speckle structured light image I l0 、I r0 And modulated multiline structured light image I l1 、I r1 、I l2 、I r2 ……I lS/W 、I rS/W . The data processing unit calculates an image I l0 、I r0 Matching cost, consistency detection and parallax optimization to obtain a parallax image ID; extracting sub-pixel coordinates of the center point of the light bar in the image; matching light bars by using the disparity map ID; and finally, generating point cloud data by using a triangular distance measurement principle, thereby completing the three-dimensional reconstruction of the measured object.
The multi-line structure projected by the invention has uniform brightness and can finish high-efficiency multi-line structure optical scanning. The invention realizes the multi-line structured light scanning of the object to be measured by combining the single-line laser with the rotary galvanometer, and the traditional scheme is to use the multi-line laser to cooperate with the rotary reflector to complete the multi-line scanning of the object to be measured. The multi-line laser consists of a laser diode, a collimating lens and a DOE module, and the multi-line laser projected by the DOE module inevitably has bright spots, which bring interference to subsequent central point extraction and generate three-dimensional reconstruction errors;
the single-line laser adopted by the invention consists of a laser diode and a Bawell prism, and the line laser converted by the point laser through the Bawell prism has the characteristics of uniform brightness and no stray light, thereby providing high-quality light bars for subsequent central point extraction. The multi-line structured light projected by the multi-line laser is reflected to the surface of an object to be measured through the rotary galvanometer, and the multi-line structured light scanning of the object to be measured is completed by controlling the rotation of the galvanometer, and because the multi-line laser is limited by the DOE module, the distance between the projected multi-line structured light is small, a large number of regions are repeatedly scanned, and the scanning efficiency is low;
the invention projects single-line laser through the single-line laser, projects the single-line laser on the surface of the measured object through the rotating prism, controls the prism to rotate and the laser to be switched on and off within the simultaneous exposure time of the left camera and the right camera, and finishes multi-line scanning of the measured object.
The camera may employ a black and white camera or a color camera.
Single line lasers, the wavelengths may be replaced with 450nm, 520nm, 658nm, or other wavelengths.
The wavelength of the laser speckle projector can be replaced by 450nm, 520nm, 658nm or other wavelengths.
The rotary galvanometer can be replaced by a plane reflecting mirror and a light splitting prism.
The control unit can be replaced by circuit boards developed by a 51-single chip microcomputer, an STM32, an auduino and the like.
The data processing unit can be replaced by CPU or GPU mini-computer such as Raspberry pie, jetson Nano, etc.
Fig. 2 is a flowchart of a binocular multiline three-dimensional reconstruction method of the present invention, and a system embodiment of the present invention includes a method embodiment of the present invention, so the method embodiment is not described herein again.
The above-mentioned embodiments are merely illustrative of the technical solutions of the present invention in a specific embodiment, and any equivalent substitutions and modifications or partial substitutions of the present invention without departing from the spirit and scope of the present invention should be covered by the claims of the present invention.
Claims (8)
1. A binocular multiline three-dimensional reconstruction method is characterized by comprising the following steps:
s1: projecting speckles to a measured object by using a laser speckle projector, and collecting a speckle image pair modulated on the surface of the measured object;
s2: performing epipolar line correction on the modulated speckle image pair and calculating matching cost to obtain a disparity map;
s3: performing multi-line structured light scanning on the object to be detected through a single-line laser and a rotary galvanometer, and sequentially acquiring a plurality of modulated multi-line structured light images on the surface of the object to be detected in the scanning process;
s4: performing epipolar line correction on a plurality of modulated multi-line structured light images, and performing sub-pixel central point extraction on light bars;
s5: carrying out light bar matching according to the disparity map, and determining the light bar matching relation in the multiple modulated multi-line structured light images;
s6: and generating point cloud data by utilizing a triangulation principle according to the sub-pixel coordinates of each central point in the light bar, and finishing the three-dimensional reconstruction of the measured object.
2. The binocular multiline three-dimensional reconstruction method according to claim 1, wherein the step S2 specifically comprises: performing epipolar line correction on the modulated speckle image pair to obtain a corrected left image and a corrected right image; creating an M-by-M region by taking an optional point in the left image as a center, searching the region with the same size in the right image along an epipolar line, and calculating the matching cost once, wherein the cost calculation formula is as follows:
C(u,v,d)＝ρ(C AD (u,v,d),λ AD )+ρ(C ce (u,v,d),λ ce )
wherein, C AD (u, v, d) is the matching cost obtained by the AD method; c ce (u, v, d) is the hamming distance based on Census transform;
is a normalized formula, where c is the matching cost value and λ is the control parameter; and obtaining the disparity map according to a cost calculation formula.
3. The binocular multiline three-dimensional reconstruction method according to claim 2, wherein the step S4 specifically comprises: performing epipolar line correction on any modulated light image with a multi-line structure, obtaining an image of light strip edges by applying a Canny operator,
performing expansion processing on the edge image and acquiring an outermost layer outline; and calculating the center point of the sub-pixel in the area in the contour by using a gray scale gravity center method, and smoothing the spline of the center point of the sub-pixel of the light bar by cubic spline interpolation.
4. The binocular multiline three-dimensional reconstruction system is characterized in that: including left camera, right camera, single line laser instrument, laser speckle projector, rotatory mirror and the control unit that shakes, the control unit controls respectively left camera, right camera, single line laser instrument, laser speckle projector and rotatory mirror that shakes, the control unit includes treater and memory, the treater processing storage is in procedure in the memory is in order to realize following step:
s1: projecting speckles to a measured object by using a laser speckle projector, and collecting a speckle image pair modulated on the surface of the measured object;
s2: performing epipolar line correction on the modulated speckle image pair and calculating matching cost to obtain a disparity map;
s3: performing multi-line structured light scanning on the object to be detected through a single-line laser and a rotary galvanometer, and sequentially acquiring a plurality of modulated multi-line structured light images on the surface of the object to be detected in the scanning process;
s4: performing polar line correction on a plurality of modulated multi-line structured light images, and performing sub-pixel central point extraction on light strips;
s5: carrying out light bar matching according to the disparity map, and determining the light bar matching relation in the multiple modulated multi-line structured light images;
s6: and generating point cloud data by utilizing a triangulation principle according to the sub-pixel coordinates of each central point in the light bar, and finishing the three-dimensional reconstruction of the measured object.
5. The binocular multiline three-dimensional reconstruction system of claim 4, wherein: the step S1 specifically comprises the following steps: the control unit controls the rotary galvanometer to move to a zero position, and when the rotary galvanometer moves to the zero position, the control unit controls to start the laser speckle projector, sends collected image pulse signals to the left camera and the right camera and obtains a speckle image pair modulated on the surface of the measured object.
6. The binocular multiline three-dimensional reconstruction system of claim 5, wherein: the step S2 specifically comprises the following steps: performing epipolar line correction on the modulated speckle image pair to obtain a corrected left image and a corrected right image; creating an M-by-M region by taking an optional point in the left image as a center, searching the region with the same size in the right image along an epipolar line, and calculating the matching cost once, wherein the cost calculation formula is as follows:
C(u,v,d)＝ρ(C AD (u,v,d),λ AD )+ρ(C ce (u,v,d),λ ce )
wherein, C AD (u, v, d) is the matching cost obtained by the AD method; c ce (u, v, d) is the hamming distance based on Census transform;
is a normalized formula, where c is the matching cost value and λ is the control parameter; and obtaining the disparity map according to a cost calculation formula.
7. The binocular multiline three-dimensional reconstruction system of claim 6, wherein: the step S3 specifically comprises the following steps: the control unit controls the single-line laser and the rotary galvanometer to perform multi-line structured light scanning on the measured object, simultaneously sends acquired image pulse signals to the left camera and the right camera, and controls the rotary galvanometer to rotate to open the single-line laser within the exposure time of the cameras and close the single-line laser after t ms; then the control unit controls the rotating position L + S of the rotary galvanometer, then the single-line laser is turned on, and the single-line laser is turned off after t ms; then the control unit controls the rotating position L +2S of the rotary galvanometer to turn on the single-line laser and turns off the single-line laser after t ms; and in the same way, obtaining a plurality of modulated multi-line structured light images, wherein L is the position of a first light strip projected, and S is the line spacing or the laser line width.
8. The binocular multiline three-dimensional reconstruction system of claim 7, wherein: the step S4 specifically comprises the following steps: performing epipolar line correction on any modulated multi-line structured light image, obtaining an optical strip edge image by applying a Canny operator, performing expansion processing on the edge image and obtaining an outermost layer profile; and calculating the center point of the sub-pixel in the area in the contour by using a gray scale gravity center method, and smoothing the spline of the center point of the sub-pixel of the light bar by cubic spline interpolation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211621678.2A CN115861542A (en) | 2022-12-17 | 2022-12-17 | Binocular multiline three-dimensional reconstruction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211621678.2A CN115861542A (en) | 2022-12-17 | 2022-12-17 | Binocular multiline three-dimensional reconstruction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115861542A true CN115861542A (en) | 2023-03-28 |
Family
ID=85673536
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211621678.2A Pending CN115861542A (en) | 2022-12-17 | 2022-12-17 | Binocular multiline three-dimensional reconstruction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115861542A (en) |
-
2022
- 2022-12-17 CN CN202211621678.2A patent/CN115861542A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3531066B1 (en) | Three-dimensional scanning method including a plurality of lasers with different wavelengths, and scanner | |
| US10810750B1 (en) | Three-dimensional scanning system and scanning method thereof | |
| CN111023970B (en) | Multi-mode three-dimensional scanning method and system | |
| US9392262B2 (en) | System and method for 3D reconstruction using multiple multi-channel cameras | |
| WO2018103693A1 (en) | Hybrid light measurement method for measuring three-dimensional profile | |
| CN106500627B (en) | 3-D scanning method and scanner containing multiple and different long wavelength lasers | |
| CN114998499B (en) | Binocular three-dimensional reconstruction method and system based on line laser galvanometer scanning | |
| US20150055853A1 (en) | Method and system for providing three-dimensional and range inter-planar estimation | |
| US11587252B2 (en) | Positioning method and system combining mark point positioning and intelligent reverse positioning | |
| US20060210146A1 (en) | Creating 3D images of objects by illuminating with infrared patterns | |
| CN110940295B (en) | High-reflection object measurement method and system based on laser speckle limit constraint projection | |
| JP2003130621A (en) | Method and system for measuring three-dimensional shape | |
| CN110702025B (en) | Grating type binocular stereoscopic vision three-dimensional measurement system and method | |
| US20220398760A1 (en) | Image processing device and three-dimensional measuring system | |
| CN108981608B (en) | Novel line structured light vision system and calibration method | |
| Lin | Automatic 3D color shape measurement system based on a stereo camera | |
| CN113160416B (en) | Speckle imaging device and method for coal flow detection | |
| JP4379626B2 (en) | Three-dimensional shape measuring method and apparatus | |
| CN114359406A (en) | Calibration of auto-focusing binocular camera, 3D vision and depth point cloud calculation method | |
| CN115861542A (en) | Binocular multiline three-dimensional reconstruction method and system | |
| RU125335U1 (en) | DEVICE FOR MONITORING LINEAR SIZES OF THREE-DIMENSIONAL OBJECTS | |
| CN212843399U (en) | Portable three-dimensional measuring equipment | |
| Abzal et al. | Development of an automatic map drawing system for ancient bas-reliefs | |
| CN115880432A (en) | Binocular multiline three-dimensional reconstruction method and system based on region coding | |
| WO2023007625A1 (en) | Three-dimensional measurement system, device, method, and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |