CN109579741A - A kind of Full-automatic multimould state three-dimensional colour measurement method based on multi-angle of view - Google Patents
A kind of Full-automatic multimould state three-dimensional colour measurement method based on multi-angle of view Download PDFInfo
- Publication number
- CN109579741A CN109579741A CN201811292271.3A CN201811292271A CN109579741A CN 109579741 A CN109579741 A CN 109579741A CN 201811292271 A CN201811292271 A CN 201811292271A CN 109579741 A CN109579741 A CN 109579741A
- Authority
- CN
- China
- Prior art keywords
- phase
- camera
- black
- phase shift
- fourier transform
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
- G01B11/25—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object
- G01B11/2545—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object with one projection direction and several detection directions, e.g. stereo
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The Full-automatic multimould state three-dimensional colour measurement method based on multi-angle of view that the invention discloses a kind of, phase shift technology of profiling is carried out using three step phase shift methods first, adding a width again using the second spoke line in three step phase shifts simultaneously, white figure carries out Fourier Transform Profilomery entirely, and phase is unfolded based on three-dimensional phase unwrapping method;Then judge the moving region of testee pixel by pixel by the variation of adjacent two moment phase;When measuring static object, high-precision profile is obtained using phase shift technology of profiling, when measuring dynamic object, efficiently extracts moving object contours using Fourier Transform Profilomery.The advantages of present invention can make full use of Fourier Transform Profilomery and phase shift technology of profiling under complex scene (not only included moving object in scene but also include stationary object or object while the two states for showing motion and standstill), obtains reconstruction result with better accuracy.
Description
Technical field
The invention belongs to 3 dimension imaging technology field, especially a kind of Full-automatic multimould state three-dimensional colour based on multi-angle of view
Measurement method.
Background technique
In high precision, untouchable 3 dimension imaging technology industrial detection, medical treatment, human-computer interaction, in terms of have
Wide application.In in the past few decades, many different three dimension profile measurement methods are suggested, such as stereo vision method,
Time-of-flight method, fringe projection method etc..The advantages of wherein fringe projection method is due to its high-precision, Yi Shixian and become now it is most main
One of three dimension profile measurement technology wanted, and with the development of electronic projection apparatus, the real time three-dimensional imaging system of high-precision, high speed
System is also possibly realized.In fringe projection technology two main methods be phase shift technology of profiling (V.Srinivasan, H.-C.Liu,
and M.Halioua,“Automated phase-measuring profilometry of 3-d diffuse
Objects, " Applied optics 23,3105-3108 (1984)) and Fourier Transform Profilomery (M.Takeda and
K.Mutoh,“Fourier transform profilometry for the automatic measurement of 3-d
object shapes,”Applied optics 22,3977-3982(1983).)。
Phase shift technology of profiling takes multiframe optical grating projection strategy to carry out three-dimensional measurement to object, more from coding mode
Frame optical grating projection strategy restores phase information in the time domain, and by shooting one group of structure light image, the light intensity sequence of acquisition is available
In unambiguously encoding each pixel, this method can effectively inhibit environment light and testee reflectivity to the shadow of phase recovery
It rings, therefore there is higher precision.But due to the characteristic of multiframe imaging, phase shift technology of profiling is more sensitive to kinematic error, with
Testee movement velocity increases, and measurement error also will increase.Error caused by reduce movement to the greatest extent to measuring, can be improved
Measurement efficiency, use bar graph reconstruction of objects three-dimensional information few as far as possible.Fourier Transform Profilomery is a kind of list based on space filtering
Frame grating project, it is only necessary to which an amplitude grating can reconstruct object dimensional information.This method is by projection sinusoidal grating, in frequency domain
Middle selection appropriate filter window extracts object phase, realizes three-dimensional reconstruction according to the mapping relations of phase and height.Due to only with one
The imaging of frame bar line, Fourier Transform Profilomery are not so good as phase shift technology of profiling to the rejection ability of environment light, therefore when measurement stationary body
Precision is not so good as phase shift technology of profiling;But to motion-insensitive, therefore in the object of measurement movement, precision is better than Fourier Transform Profilomery
Phase shift technology of profiling.
For the contradiction between above-mentioned measurement efficiency and measurement accuracy, solving the contradictory method is to reduce phase shift profile
Art is to the sensitivity of movement or the precision of promotion Fourier Transform Profilomery.Reducing phase shift technology of profiling can to the sensitivity of movement
Set about in terms of three: (1) improving the speed of hardware device;(2) efficiency of phase unwrapping is improved;(3) phase shift technology of profiling is removed
Error caused by middle movement.The method for promoting Fourier Transform Profilomery precision is mainly to remove the zero-frequency information for representing background to carrying
The influence of the fundamental frequency of object phase information.
Although these above-mentioned methods improve the dynamic measurement capability of phase shift technology of profiling to a certain extent, improve in Fu
The precision of leaf technology of profiling, but (scene includes simultaneously moving object and stationary object or the same object is in complex scene
Existing different motion feature) under still seldom go out a higher reconstruction result of precision.
Summary of the invention
The Full-automatic multimould state three-dimensional colour measurement method based on multi-angle of view that the purpose of the present invention is to provide a kind of, passes through
To the accurate judgement of testee motion state, it is static schema when measuring stationary body, is obtained using phase shift technology of profiling
High-precision profile;It is dynamic mode when measuring dynamic object, efficiently extracts moving object wheel using Fourier Transform Profilomery
Exterior feature, and the switching between mode be it is full automatic, do not need any human intervention.
The technical solution for realizing the aim of the invention is as follows: a kind of Full-automatic multimould state three-dimensional colour survey based on multi-angle of view
Amount method, it is characterised in that steps are as follows:
Step 1 adjusts the spatial position of four cameras and a projector in measuring system, completes the calibration of system;
Step 2 obtains measured object wrapped phase with Fourier Transform Profilomery and phase shift technology of profiling respectively, utilizes three-dimensional phase
Phase is unfolded in the method for development, finds out the first black and white camera i.e. main phase machine under two methods of Fourier Transform Profilomery and phase shift technology of profiling respectively
Sub-pix match point;
Step 3 chooses suitable threshold using the absolute phase difference at current time and last moment Fourier Transform Profilomery
Value judges moving region, to object of which movement region Fourier Transform Profilomery as a result, stagnant zone phase shift technology of profiling result;
Step 4, the space three-dimensional point of measured object is calculated with the sub-pix match point of fusion, and obtains color camera in master
The color texture of corresponding points on camera, the true color for completing three-dimensional data are shown.
Compared with prior art, the present invention its remarkable advantage: (1) and tradition only uses phase shift technology of profiling or Fourier Transform Profilomery
System compare, this method can obtain the result for having higher precision under compound movement scene;(2) with improve hardware device speed
Method compare, this method does not need expensive high speed acquisition and projection device, and cost is relatively low;(3) it and eliminates caused by movement
The method of error is compared in phase shift technology of profiling, and the algorithm is simple, calculation amount is small, can be realized real-time online measuring.
Present invention is further described in detail with reference to the accompanying drawing.
Detailed description of the invention
Fig. 1 is the Full-automatic multimould state three-dimensional colour measurement method flow diagram based on multi-angle of view.
Fig. 2 is four mesh fringe projection system organigrams.
Fig. 3 is the measurement result under first complex scene, and (a), (b) are respectively the knot measured under traditional three step phase shift methods
It is that fruit and Full-automatic multimould state three-dimensional colour measurement method proposed by the present invention measure as a result, (c), (g) be respectively traditional three steps
It is measured under error, proposition method of the present invention between the spherical surface result measured under phase shift method and the spherical surface being fitted with the result flat
Error between hardened fruit and the plane being fitted with the result, (d), (h) respectively indicate the histogram of (c), (d);(e), (i) point
The plane result that Wei not measure under traditional three step phase shift methods and with the error between the plane of result fitting, moving party of the present invention
Error between the plane result that is measured under method and the plane being fitted with the result, (f), (j) respectively indicate the histogram of (e), (i)
Figure.
Fig. 4 is the measurement result under second complex scene, wherein (a), (b) are respectively to measure under traditional three step phase shift methods
Result and Full-automatic multimould state three-dimensional colour measuring system proposed by the present invention measure as a result, (c), (d) be respectively tradition
It is measured under error, proposition method of the present invention between the plate result measured under three step phase shift methods and the plane being fitted with the result
Plate result and the plane that is fitted with the result between error, (e), (f) respectively indicate the histogram of (c), (d).
Fig. 5 be real-time scene under measurement result, wherein (a) be small fan measurement scene, (b), (c) be respectively this hair
In the movement that the Full-automatic multimould state three-dimensional colour measuring system of bright proposition measures small fan as a result, (d) be David measurement
Scene, (e), (f) be respectively David that Full-automatic multimould state three-dimensional colour measuring system proposed by the present invention measures result.
Specific embodiment
The present invention by the variation of adjacent two moment object phase information come judgment object moving region pixel by pixel, for
Implement the fusion of Fourier Transform Profilomery and phase shift technology of profiling in the region of object different motion state.
In conjunction with Fig. 1, the present invention is based on the realization of the Full-automatic multimould state three-dimensional colour measurement method of multi-angle of view, steps are as follows:
Step 1, building measuring system, (also known as four mesh fringe projection systems are to complete four mesh fringe projections and survey
Amount), and complete system calibrating.Four mesh fringe projection systems include a computer, three black and white cameras, a color camera,
One projector is connected between projector and four cameras with four firing lines, four radical evidences of four cameras and computer
Line is connected.Detailed process is as follows: firstly, since the measuring system built is that (system of the camera quantity greater than 1 claims more mesh systems
For more mesh systems), the position of camera, projector in system need to be optimized, the method for optimization are as follows: the distance between visual angle
It is closer, then Projection Line Segment of any point on a visual angle within the scope of the certain depth on another visual angle is shorter, separately
On the one hand the distance between two visual angles are bigger, then between two visual angles three-dimensionalreconstruction it is more accurate (T.Tao, Q.Chen,
S.Feng,Y.Hu,M.Zhang,and C.Zuo,"High-precision real-time 3D shape measurement
Based on a quad-camera system, " 20,014009 (2018) of Journal of Optics), system after optimization
The placement position of each component are as follows: the first black and white camera 1 keeps more remote with projector, i.e. the first black and white camera 1 and projector
Between reserve the gap that can put a camera again, the second black and white camera 2 is placed between the first black and white camera 1 and projector simultaneously
And put close to the first black and white camera 1, third black and white camera 4 is symmetrically put with the first black and white camera 1 about projector, colored phase
Machine 3 and the second black and white camera 2 are symmetrically placed about projector, and each component is put as shown in Figure 2.
Then Zhang Zhengyou calibration algorithm (Z.Zhang, " A flexible new technique for camera are utilized
calibration.”IEEE Transactions on pattern analysis and machine
Intelligence.22 (11), 1330-1334 (2000)) by under whole system calibration to unified world coordinate system, obtain four
The calibrating parameters of a camera and projector under a world coordinate system, and 2 d-to-3 d is converted by these parameters, two dimension
To two-dimensional mapping parameters (K.Liu, Y.Wang, D.L.Lau, et al, " Dual-frequency pattern scheme
for high-speed 3-D shape measurement.”Optics express.18(5):5229-5244(2010).)。
Step 2 obtains measured object wrapped phase with Fourier Transform Profilomery and phase shift technology of profiling respectively, utilizes three-dimensional phase
Phase (T.Tao, Q.Chen, S.Feng, Y.Hu, M.Zhang, and C.Zuo, " High-precision is unfolded in the method for development
real-time 3D shape measurement based on a quad-camera system,"Journal of
20,014009 (2018) of Optics), the first black and white phase under two methods of Fourier Transform Profilomery and phase shift technology of profiling is found out respectively
The sub-pix match point of machine 1, specific steps are as follows: throw the stripe pattern of 1+3 with projector first, i.e. white figure adds three three to a width entirely
The image for walking phase shift, the image deformed by four camera synchronous acquisitions, then uses phase shift technology of profiling according to the image of acquisition respectively
Solve wrapped phase with Fourier Transform Profilomery, solution procedure is as follows: for phase shift technology of profiling, the first black and white camera 1 is collected
Three step phase shifted images (being illustrated by taking the first black and white camera 1 as an example below) can indicate are as follows:
In above formula, (uC,vC) indicate camera on a pixel coordinate,Indicate camera collected three
Width plot of light intensity, ACIndicate average intensity, BCIndicate modulation degree light intensity, ΦCIndicate the absolute phase of bar graph.
If the striped of projection is nicking, a little correspondence on projector can be calculated in camera according to the following formula
Point abscissa:
U in formulaP(uC,vC) indicate camera image on pixel corresponding points on projector abscissa, WPIndicate projection
The lateral resolution of instrument, N indicate the radical of striped.Then the coordinate of 3D point can be found out by following three formula:
M in formulaCP、NCP、CCP、It indicates to use camera and the resulting parameter matrix of projector calibrating,
XW、YW、ZWIndicate that the coordinate of the 3D point found out, the above process can be used between the first black and white camera 1 and other two black and white cameras
Reconstruct, the point in the first black and white camera 1 can be mapped in another camera by calibrating parameters, be looked in another camera
To the sub-pix match point in the first black and white camera 1, to complete the reconstruct between two cameras.But due to arctan function
Truncation effect can only obtain wrapped phase using collected bar graph:
φ in formulaC(uC,vC) indicate the wrapped phase solved.The relationship of wrapped phase and absolute phase are as follows:
ΦC(uC,vC)=φC(uC,vC)+2kC(uC,vC)π,kC(uC,vC)∈[0,N-1]
K in formulaCIndicate that fringe order, N indicate the radical of striped.The process for solving fringe order is just called unpacking.It is right
In Fourier Transform Profilomery, the present invention (selects the second width bar graph using the second width bar graph in complete white figure and three step phase shift figures
The reason of will be addressed below) Lai Jinhang Fourier Transform Profilomery and find out the wrapped phase under Fourier Transform Profilomery.First is black
The collected Quan Baitu of Bai Xiangji 1It may be expressed as:
For the precision for improving Fourier Transform Profilomery, the present invention uses background normalization Fourier Transform Profilomery, background normalization
Be first subtract Quan Baitu with bar graph to eliminate the DC component in frequency spectrum, then divided by the sum of informal voucher line and a small constant with
The influence of reflectivity is eliminated, such as following formula:
In formulaStriped light intensity after indicating background normalization, b indicate the constant of a very little, and it is anti-for adding a small constant
Only there is for 0 the case where in denominator.Can be eliminated using background normalization method become since testee edge is sharp, surface is discontinuous and table
Spectral aliasing caused by phenomena such as face reflectivity changes are big influences, while ensure that measurement accuracy.By the item after background normalization
Line figure carries out Fourier transformation and obtains spectrogram, and extracting level-one frequency spectrum with Hanning window body and doing inverse Fourier transform again just can be obtained
Wrapped phase under Fourier Transform Profilomery.
Then unpacking is carried out to obtain absolute phase, for the effect for improving unpacking to the wrapped phase under two methods
Rate, the present invention is using three-dimensional phase unwrapping method (T.Tao, Q.Chen, S.Feng, Y.Hu, M.Zhang, and C.Zuo, " High-
precision real-time 3D shape measurement based on a quad-camera system,"
20,014009 (2018) of Journal of Optics) carry out unpacking, and to reduce calculation amount, only Fourier Transform Profilomery is asked
The wrapped phase obtained carries out three-dimensional phase unwrapping, the package phase being then unfolded under phase shift technology of profiling with the absolute phase of expansion
Position.Specific expansion process is as follows: the fringe order at any point has N kind possible in the first black and white camera 1, that is, has N kind may
Absolute phase, corresponding N number of three-dimensional candidate point can be solved using calibrating parameters.Due to the limit of projector in measuring system
System, effective drop shadow spread (guaranteeing the sufficiently large range of the fringe contrast that projector is launched) are not infinitely great, effective throwings
Shadow range also limits the areas imaging of fringe projection system, obtains system according to effective drop shadow spread of projector in system
Areas imaging first excludes the three-dimensional candidate point outside the range, so that the quantity of three-dimensional candidate point is reduced to N1It is a, it will
This N1A three-dimensional candidate point, which projects in the second black and white camera 2, obtains corresponding two-dimension candidate point, this each two-dimension candidate point
There is a corresponding wrapped phase, phase portrait inspection is carried out to these wrapped phases: by these wrapped phases and first
The wrapped phase at 1 midpoint of black and white camera makees difference and takes absolute value and a threshold value comparison, and threshold value is selected as 0.6rad here, by it
In be greater than threshold value two-dimension candidate point exclude.It in this way can be by the quantity of two-dimension candidate point and three-dimensional candidate point from N1Drop to N2
It is a, N2A three-dimensional candidate point is using obtaining N on polar curve mapping projections to third black and white camera 42A two-dimension candidate point, then to this N2
A two-dimension candidate point carries out phase portrait inspection, and the threshold value chosen here is 1rad, can be by the quantity of three-dimensional candidate point into one
Step is reduced to N3, N above1、N2、N3Meet following relationship:
1≤N3≤N2≤N1≤N
For in the first black and white camera 1 it is arbitrary a bit, corresponding N3Not necessarily 1 (ideally N3It is equal to
1), that is, fringe order cannot be uniquely determined.In order in remaining N3A candidate point, this hair are uniquely determined in a candidate point
It is bright to use the sum of weight phase difference algorithm: for N3Each of a candidate point candidate point, passes through following formula:
Make full use of twice phase portrait examine obtained phase difference, wherein Δ φ be phase difference weight it
With,For phase portrait twice examine phase difference,Corresponding to phase differenceWeight factor.Subpoint precision is higher, and phase difference reliability is higher, ought to distribute bigger power
Repeated factor, therefore the value of weight factor in the present invention is as follows:It is final to choose the lesser time of Δ φ
Reconnaissance is correct corresponding points, can thus uniquely determine a fringe order, and the wrapped phase of Fourier Transform Profilomery is complete
Expansion.After expansion, then pass through the wrapped phase of following formula expansion phase shift technology of profiling:
In formulaIndicate the absolute phase of Fourier Transform Profilomery, round indicates round operation.?
After absolute phase under to two methods, is obtained further according to the main phase machine i.e. calibrating parameters of the first black and white camera 1 and projector
Then three-dimensional point is mapped in third black and white camera 4 using polar curve mapping and is obtained by corresponding three-dimensional point under one black and white camera, 1 visual angle
To corresponding points of first black and white camera 1 in third black and white camera 4, Fourier Transform Profilomery and phase shift are finally found by interpolation arithmetic
Sub-pix match point under technology of profiling.
Step 3 chooses suitable threshold using the absolute phase difference at current time and last moment Fourier Transform Profilomery
Value judges moving region, to the moving region Fourier Transform Profilomery of object as a result, stagnant zone phase shift technology of profiling
As a result, specific steps are as follows: judge to move using phase frame difference method, i.e., judged with the difference degree of adjacent moment two phase place each
The motion conditions of point.First against selection any two phases come the problem of comparison, the present invention is selected in Fu with adjacent two moment
The phase of leaf technology of profiling is compared, the process of selection and the reason is as follows that: firstly, phase and phase that Fourier Transform Profilomery acquires
It moves the phase that technology of profiling acquires and cannot bring and compare, because Fourier Transform Profilomery is to extract phase letter in a frequency domain with filter window
Breath, the result that this method obtains is smoother relative to the result of phase shift technology of profiling, therefore is also under testee static position
There is certain deviation with phase shift technology of profiling;Compare secondly, the phase merged can not be brought, because also will appear Fu in this way
In the phase of leaf technology of profiling the case where comparing with the phase of phase shift technology of profiling;Only it is left the phase with two phase shift technology of profiling in this way
Position is come the scheme that relatively or with the phase of two Fourier Transform Profilomeries compares.But under moving scene, phase shift technology of profiling
The phase found out is the phase of the inaccuracy with movement ripple, and to the phase that the Fourier Transform Profilomery of motion-insensitive is found out
It is then accurate phase, so the phase of final choice Fourier Transform Profilomery carries out phase frame difference method.
Followed by motion determination, makees difference with the phase of the Fourier Transform Profilomery of adjacent two frame and take absolute value and one
Threshold value comparison, if absolute value is greater than the threshold value, then it is assumed that point movement, it is on the contrary then think that the point is static.For accurate judgment object
Motion state, need to select reasonable threshold value, distinguish the case where noise causes and the case where movement causes.The process that threshold value is chosen
It is as follows: firstly, to n continuous moment t under static scene0~tn-1Fourier Transform Profilomery phase diagram in some fixed area
The phase in domain is averaging to obtain an average phase, the average phase acquired can approximate representation be under the influence of no ambient light noise
Phase, with the average phase and tn-1The phase at moment makees difference and seeks its histogram, weeds out the histogram area counted less than 50
Then domain obtains the maximum section of difference.Absolute value is asked to the region, takes its upper limit as phase threshold, which is quiet
The phase maximum fluctuation due to caused by the factors such as ambient light interference in the case of only, as long as certain point phase fluctuation is greater than under static position
Maximum fluctuation, then it is assumed that the point movement.Final deterministic process is as follows: the phase difference ΔΦ of t moment and t- time ΔtCIt can table
It is shown as:
In formulaIndicate that t moment phase changes with time rate, Δ t indicates the difference at adjacent two moment.?
After acquiring threshold value with above-mentioned Research on threshold selection, if certain point phase variation be greater than threshold value, then it is assumed that the point movement, it is on the contrary then
Think that the point is static:
In formula, ΔΦCIndicate the phase difference at adjacent two moment, threΦIndicate the threshold value of judgement,Table
Show that t moment judges the output of movement as a result, result indicates movement for 1,0 indicates static.
Then it is directed to static point, uses the sub-pix match point of phase shift technology of profiling;For the point of movement, Fourier is used
The sub-pix match point of technology of profiling.In order to guarantee the precision of final fusion results, result and phase in Fourier Transform Profilomery are avoided
The phenomenon that being truncated is moved at the fusion of the result of technology of profiling, Fourier Transform Profilomery result and phase under different motion state are analyzed
The difference for moving technology of profiling result, needs to reasonably select the width in three step phase shift stripe patterns to do Fourier transformation, selection
Process is as follows: the bar graph in the collected movement of the first black and white camera 1 may be expressed as:
In formulaIndicate three width plot of light intensity in the collected movement of camera,Respectively indicate fortune
It moves to phase displacement error caused by the first width and third spoke line.For the object of translational motion,Meet with ShiShimonoseki
System:
Or
Under the motion conditions, finally with Fourier's wheel that the phase that phase shift technology of profiling solves is along second raster pattern
The result of wide art fluctuates up and down.
For the object of rotary motion, translational motion is also served as to handle, if continuous rotary motion is regarded as multiple
Object then can be divided into two parts along rotation center by the discrete motion in short time interval, and a part is to do in one direction
Translational motion, another part are to do translational motion towards another direction.
The object moved in the opposite direction for changing the direction of motion suddenly, if one group of the acquisition of the first black and white camera 1
In three step phase-shifted grating figures, the lucky existing raster pattern moved along first direction, and have the grating moved along second direction
Figure, thenMeet following relationship:
Or
Under the motion state, it is final with three step phase shift methods acquire the result is that in Fourier's result along third width figure
Lower fluctuation, but object changes direction and only occurs in a flash, and the first black and white camera 1 not necessarily collects two fortune simultaneously
Raster pattern under dynamic direction, so can not consider the motion state.To sum up, it should finally select to be in Fu of the second amplitude grating figure
Leaf variation, and the result for the moving region and phase shift technology of profiling judged according to phase frame difference method is merged.
Step 4, the space three-dimensional point of measured object is calculated with the sub-pix match point of fusion, and obtains the colored phase of third
The color texture of the corresponding points in the first black and white camera 1 of machine 3, the true color for completing three-dimensional data are shown.It is specific as follows: to obtain and melt
After sub-pix match point after conjunction, three of the reconstruct with the first black and white camera 1 and third black and white camera 4 can be acquired according to the following formula
Dimension data:
In formulaIndicate the three-dimensional data reconstructed, AndFor reflecting by the 2 d-to-3 d between the first black and white camera 1 obtained by calibrating and third black and white camera 4
Penetrate parameter.
Correspondence of first black and white camera 1 on third color camera 3 can be solved according to obtained three-dimensional data
Point obtains the color texture of corresponding points, completes three-dimensional data true color and shows, logical for any point in the first black and white camera 1
It crosses following formula and acquires its corresponding points on color camera:
In formulaIndicate corresponding points of the first black and white camera 1 acquired on third color camera 3,WithIt, will for the calibrating parameters matrix in step 1The colour information at place is mapped toOn just can
Three-dimensional data true color is completed to show.
Embodiment
For the feasibility and real-time for testing the method for the invention, a set of suitable dynamic scene has been built using the present invention
With four mesh colour real time three-dimensional imaging systems of static scene, the system structure is as shown in Figure 2.Within the system, three have been used
Black and white camera carries out three-dimensional phase unwrapping, pastes color texture using a color camera, first in three black and white cameras such as Fig. 2
Black and white camera 1, the second black and white camera 2 and third black and white camera 4, model Basler acA640-750um, color camera is as schemed
In third color camera 3, model Basler acA640-750uc, highest frame per second is all 750fps, maximum resolution
It is 640 × 480, all employs the Computar camera lens of 12mm, the projector model LightCrafter 4500Pro used,
Resolution ratio is 912 × 1140, maximum speed 120Hz.In an experiment, the speed of projector is 100Hz, and all cameras are all
It is triggered with projector, the setting of relevant parameter are as follows: judge the phase threshold of movement
Two complicated measurement scenes are devised first.First is scene that existing moving object has stationary object again,
There is the precision ball of static plate and movement under the scene.Measurement result is as shown in figure 3, (a), (b) are respectively traditional three step phases
It is that the result and Full-automatic multimould state three-dimensional colour measurement method proposed by the present invention measured under shifting method measures as a result, (c), (g)
Error, the present invention between the spherical surface result measured under respectively traditional three step phase shift methods and the spherical surface being fitted with the result propose
Error between the plate result that is measured under method and the plane being fitted with the result, (d), (h) respectively indicate (c), (d) it is straight
Fang Tu;(e), (i) is respectively the plane result measured under traditional three step phase shift methods and with the mistake between the plane of result fitting
Difference, the plane result measured under proposition method of the present invention and with the error between the plane of result fitting, (f), (j) distinguish table
Show the histogram of (e), (i).It can be seen that from measurement result, for static plate, the result of final fusion method is almost
The precision of phase shift technology of profiling is remained, precision is 37 μm;The detectable movement out of method for the precision ball of movement, in the present invention
Precision ball and by the result of phase shift technology of profiling be substituted for Fourier Transform Profilomery as a result, under traditional three step phase shift methods, measurement
Precision is 68 μm, and the precision of the method for the present invention is 44 μm.
Second complex scene is object while being in static and two states of movement scene.We measure one and do
The plate of rotary motion, plate are believed that at rotation center and remain static, and are in movement shape on the both sides of rotation center
State.Last measurement result is as shown in Figure 4.Wherein (a), (b) are respectively the result and the present invention measured under traditional three step phase shift methods
It is being measured under the phase shift of proposition and Fourier's fusion method as a result, (c), (d) be respectively the plate measured under traditional three step phase shift methods
As a result the error between plane that and with the result is fitted, the plate result measured under proposition method of the present invention and intended with the result
Error between the plane of conjunction, (e), (f) respectively indicate the histogram of (c), (d).It can be seen that from measurement result, rotary motion
Under, plate center part be phase shift technology of profiling as a result, both sides used Fourier Transform Profilomery as a result, in traditional phase
The error of fitting on plate both sides is very big under technology of profiling, and precision is 77.7 μm, under the method according to the invention, in phase shift technology of profiling and Fu
There is not rank truncation phenomenon in the part of leaf technology of profiling fusion, and final precision is 39.8 μm.Measurement knot under two above scene
Fruit demonstrates method proposed by the present invention can obtain the higher reconstruction result of precision under complex scene.
Fan and David's model in the movement of last real-time measurement, measurement result such as Fig. 5.Wherein (a) is small fan
Measure scene, (b), (c) be respectively in the movement that is measured under phase shift technology of profiling and fusion method small fan as a result, (d) being big
The measurement scene defended, (e), (f) be respectively the David's as a result, from measurement result measured under phase shift technology of profiling and fusion method
It can be observed, in case of motion, phase shift technology of profiling has apparent movement ripple, merges rule well by phase shift profile
The measurement result of art is replaced into the result of Fourier Transform Profilomery.Final this system is able to achieve the measuring speed of 25fps.
Claims (6)
1. a kind of Full-automatic multimould state three-dimensional colour measurement method based on multi-angle of view, it is characterised in that steps are as follows:
Step 1 adjusts the spatial position of four cameras and a projector in measuring system, completes the calibration of system;
Step 2 obtains measured object wrapped phase with Fourier Transform Profilomery and phase shift technology of profiling respectively, utilizes three-dimensional phase unwrapping
Phase is unfolded in method, finds out the first black and white camera i.e. Asia of main phase machine under two methods of Fourier Transform Profilomery and phase shift technology of profiling respectively
Pixel matching point;
Step 3 is chosen suitable threshold value and is sentenced using the absolute phase difference at current time and last moment Fourier Transform Profilomery
Disconnected moving region out, to object of which movement region Fourier Transform Profilomery as a result, stagnant zone phase shift technology of profiling result;
Step 4, the space three-dimensional point of measured object is calculated with the sub-pix match point of fusion, and obtains color camera in main phase machine
The color texture of upper corresponding points, the true color for completing three-dimensional data are shown.
2. according to the method described in claim 1, it is characterized in that in step 1, the placement position of each component of measuring system are as follows:
The gap for putting a camera again is reserved between first black and white camera and projector, the second black and white camera is placed in the first black and white camera
It is put between projector and close to the first black and white camera, third black and white camera is symmetrical about projector with the first black and white camera
It puts, color camera is symmetrically placed about projector with the second black and white camera, is triggered between projector and four cameras with four
Line is connected, and four cameras are connected with computer with four data lines.
3. according to the method described in claim 1, it is characterized in that carry out phase shift technology of profiling with three step phase shift methods in step 2,
Add the Fourier Transform Profilomery of a width white figure progress background normalization entirely again using the second spoke line in three step phase shift methods simultaneously, the
The collected four width figure representation of one black and white camera are as follows:
In above formula, (uC,vC) indicate camera on a pixel coordinate,Indicate the collected Quan Baitu of camera,Indicate the collected three width plot of light intensity of camera, ACIndicate average intensity, BCIndicate modulation degree light intensity, ΦCIndicate item
The phase of line figure.
4. according to the method described in claim 1, it is characterized in that judging to move by phase frame difference method, i.e., in step 3
The motion conditions of a point are judged with the difference degree of adjacent moment two phase place, it is specific as follows: the phase of t moment and t- time Δt
Potential difference ΔΦCIt indicates are as follows:
In formulaIndicate that t moment phase changes with time rate, as long as variation of certain point phase is greater than threshold value,
Think that the point moves, specific deterministic process is as follows:
In formula, threΦIndicate the threshold value of judgement,Indicate that t moment judge the output moved as a result, result is 1
Indicate movement, 0 indicates static.
5. according to the method described in claim 3, it is characterized in that it is as follows that threshold value chooses process: firstly, in static state in step 3
To n continuous moment t under scene0~tn-1Fourier Transform Profilomery phase diagram in the phase of some fixed area be averaging
To an average phase, the average phase that acquires can approximate representation be phase under the influence of no ambient light noise, it is average with this
Phase and tn-1The phase at moment makees difference and seeks its histogram, weeds out the histogram regions counted less than 50, then obtains two
Absolute value is sought to the region in the maximum section of difference, takes its upper limit as phase threshold, the threshold value be under static position due to
Phase maximum fluctuation caused by ambient light interference factor, as long as certain point phase fluctuation is greater than the maximum fluctuation under static position,
Think that the point moves.
6. according to the method described in claim 1, it is characterized in that being reasonably selected in three step phase shift stripe patterns in step 3
One width does Fourier transformation, and the process of selection is as follows: the collected bar graph of the first black and white camera may be expressed as:
In formulaIndicate three width plot of light intensity in the collected movement of camera,Respectively indicate movement pair
Phase displacement error caused by first width and third spoke line;For the object of translational motion,Meet following relationship:
Or
Under the motion conditions, finally with the Fourier Transform Profilomery that the phase that phase shift technology of profiling solves is along the second amplitude grating figure
Result up and down fluctuate;
It for the object of rotary motion, is handled as translational motion, if continuous rotary motion was regarded as between multiple short time
Every interior discrete motion, then object is divided into two parts along rotation center, a part is to do translational motion in one direction, separately
A part is to do translational motion towards another direction;
The object moved in the opposite direction for changing the direction of motion suddenly, if one group of three step phase of the first black and white camera acquisition
In shifted raster figure, the lucky existing raster pattern moved along first direction, and have the raster pattern moved along second direction, thenMeet following relationship:
Or
Under the motion state, it is final with three step phase shift methods acquire the result is that along wave above and below Fourier's result of third width figure
It is dynamic, it should finally select to do Fourier's variation with the second amplitude grating figure.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811292271.3A CN109579741B (en) | 2018-11-01 | 2018-11-01 | Full-automatic multi-mode three-dimensional color measurement method based on multiple visual angles |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811292271.3A CN109579741B (en) | 2018-11-01 | 2018-11-01 | Full-automatic multi-mode three-dimensional color measurement method based on multiple visual angles |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109579741A true CN109579741A (en) | 2019-04-05 |
CN109579741B CN109579741B (en) | 2020-06-26 |
Family
ID=65921068
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811292271.3A Active CN109579741B (en) | 2018-11-01 | 2018-11-01 | Full-automatic multi-mode three-dimensional color measurement method based on multiple visual angles |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109579741B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111337517A (en) * | 2020-03-09 | 2020-06-26 | 广西师范大学 | Cylindrical lens defect detection device and cylindrical lens defect detection method |
CN112001888A (en) * | 2020-07-21 | 2020-11-27 | 浙江大学 | Impeller running state identification method and device |
CN112070842A (en) * | 2020-07-28 | 2020-12-11 | 安徽农业大学 | Multi-camera global calibration method based on orthogonal coding stripes |
CN112325799A (en) * | 2021-01-07 | 2021-02-05 | 南京理工大学智能计算成像研究院有限公司 | High-precision three-dimensional face measurement method based on near-infrared light projection |
WO2021027719A1 (en) * | 2019-08-09 | 2021-02-18 | 南京理工大学 | Reflector-based calibration method for fringe projection system |
CN112700484A (en) * | 2020-12-31 | 2021-04-23 | 南京理工大学智能计算成像研究院有限公司 | Depth map colorization method based on monocular depth camera |
CN114279360A (en) * | 2021-12-27 | 2022-04-05 | 天津大学 | Multi-eye phase deflection measurement method and device based on telecentric imaging system |
CN114459384A (en) * | 2022-02-28 | 2022-05-10 | 嘉兴市像景智能装备有限公司 | Phase shift profilometry based on multi-angle sine stripe light field fusion |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1554926A (en) * | 2003-12-26 | 2004-12-15 | 四川大学 | Method for realizing high speed moving article 3D profile measurement using stroboscopic structure lighting |
WO2007109127A3 (en) * | 2006-03-16 | 2008-04-24 | Univ Duke | Methods, systems, and computer program products for performng real-time quadrature projection based fourier domain optical coherence tomography |
CN101441066A (en) * | 2008-12-23 | 2009-05-27 | 西安交通大学 | Phase de-packaging method of color fringe coding |
CN103791856A (en) * | 2014-01-21 | 2014-05-14 | 南京理工大学 | Phase solving and unwrapping method based on four-width optical grating stripe image |
CN105066906A (en) * | 2015-07-24 | 2015-11-18 | 南京理工大学 | Fast high dynamic range three-dimensional measurement method |
CN106091981A (en) * | 2016-05-27 | 2016-11-09 | 西安交通大学 | Region projection method for measuring optical three-dimensional contour for HDR object |
CN106931910A (en) * | 2017-03-24 | 2017-07-07 | 南京理工大学 | A kind of efficient acquiring three-dimensional images method based on multi-modal composite coding and epipolar-line constraint |
CN107044833A (en) * | 2017-03-24 | 2017-08-15 | 南京理工大学 | A kind of ultrafast 3 D measuring method and its system based on improved Fourier transform outline technology |
CN108036740A (en) * | 2017-12-05 | 2018-05-15 | 南京理工大学 | Three-dimensional colour measuring system and method during a kind of high-precision real based on various visual angles |
CN108596008A (en) * | 2017-12-12 | 2018-09-28 | 南京理工大学 | The facial jitter compensation method measured for three-dimensional face |
-
2018
- 2018-11-01 CN CN201811292271.3A patent/CN109579741B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1554926A (en) * | 2003-12-26 | 2004-12-15 | 四川大学 | Method for realizing high speed moving article 3D profile measurement using stroboscopic structure lighting |
WO2007109127A3 (en) * | 2006-03-16 | 2008-04-24 | Univ Duke | Methods, systems, and computer program products for performng real-time quadrature projection based fourier domain optical coherence tomography |
CN101441066A (en) * | 2008-12-23 | 2009-05-27 | 西安交通大学 | Phase de-packaging method of color fringe coding |
CN103791856A (en) * | 2014-01-21 | 2014-05-14 | 南京理工大学 | Phase solving and unwrapping method based on four-width optical grating stripe image |
CN105066906A (en) * | 2015-07-24 | 2015-11-18 | 南京理工大学 | Fast high dynamic range three-dimensional measurement method |
CN106091981A (en) * | 2016-05-27 | 2016-11-09 | 西安交通大学 | Region projection method for measuring optical three-dimensional contour for HDR object |
CN106931910A (en) * | 2017-03-24 | 2017-07-07 | 南京理工大学 | A kind of efficient acquiring three-dimensional images method based on multi-modal composite coding and epipolar-line constraint |
CN107044833A (en) * | 2017-03-24 | 2017-08-15 | 南京理工大学 | A kind of ultrafast 3 D measuring method and its system based on improved Fourier transform outline technology |
CN108036740A (en) * | 2017-12-05 | 2018-05-15 | 南京理工大学 | Three-dimensional colour measuring system and method during a kind of high-precision real based on various visual angles |
CN108596008A (en) * | 2017-12-12 | 2018-09-28 | 南京理工大学 | The facial jitter compensation method measured for three-dimensional face |
Non-Patent Citations (5)
Title |
---|
BEIWEN LI等: "Motion-induced error reduction by combining Fourier transform profilometry with phase-shifting profilometry", 《OPTICS EXPRESS》 * |
JIAMING QIAN等: "Motion-artifact-free dynamic 3D shape measurement with hybrid Fourier-transform phase-shifting profilometry", 《OPTICS EXPRESS》 * |
TIANYANG TAO等: "High-precision real-time 3D shape measurement based on a quad-camera system", 《JOURNAL OF OPTICS》 * |
张晓璇等: "一种两步相移相位解算方法", 《光子学报》 * |
笪健等: "结合极线约束和散斑相关的实时三维测量方法", 《光学学报》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021027719A1 (en) * | 2019-08-09 | 2021-02-18 | 南京理工大学 | Reflector-based calibration method for fringe projection system |
US11808564B2 (en) | 2019-08-09 | 2023-11-07 | Nanjing University Of Science And Technology | Calibration method for fringe projection systems based on plane mirrors |
CN111337517A (en) * | 2020-03-09 | 2020-06-26 | 广西师范大学 | Cylindrical lens defect detection device and cylindrical lens defect detection method |
CN112001888A (en) * | 2020-07-21 | 2020-11-27 | 浙江大学 | Impeller running state identification method and device |
CN112001888B (en) * | 2020-07-21 | 2023-12-29 | 浙江大学 | Impeller running state identification method and device |
CN112070842A (en) * | 2020-07-28 | 2020-12-11 | 安徽农业大学 | Multi-camera global calibration method based on orthogonal coding stripes |
CN112070842B (en) * | 2020-07-28 | 2023-03-21 | 安徽农业大学 | Multi-camera global calibration method based on orthogonal coding stripes |
CN112700484A (en) * | 2020-12-31 | 2021-04-23 | 南京理工大学智能计算成像研究院有限公司 | Depth map colorization method based on monocular depth camera |
CN112325799A (en) * | 2021-01-07 | 2021-02-05 | 南京理工大学智能计算成像研究院有限公司 | High-precision three-dimensional face measurement method based on near-infrared light projection |
CN114279360A (en) * | 2021-12-27 | 2022-04-05 | 天津大学 | Multi-eye phase deflection measurement method and device based on telecentric imaging system |
CN114279360B (en) * | 2021-12-27 | 2023-08-11 | 天津大学 | Method and device for measuring multi-order phase deflection based on telecentric imaging system |
CN114459384A (en) * | 2022-02-28 | 2022-05-10 | 嘉兴市像景智能装备有限公司 | Phase shift profilometry based on multi-angle sine stripe light field fusion |
Also Published As
Publication number | Publication date |
---|---|
CN109579741B (en) | 2020-06-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109579741A (en) | A kind of Full-automatic multimould state three-dimensional colour measurement method based on multi-angle of view | |
Zennaro et al. | Performance evaluation of the 1st and 2nd generation Kinect for multimedia applications | |
CN101694375B (en) | Stereoscopic vision detecting method for measuring three-dimensional morphology on strong reflection surface | |
Wang et al. | Robust active stereo vision using Kullback-Leibler divergence | |
Gorthi et al. | Fringe projection techniques: whither we are? | |
Sun et al. | Object surface recovery using a multi-light photometric stereo technique for non-Lambertian surfaces subject to shadows and specularities | |
CN107607040A (en) | A kind of three-dimensional scanning measurement device and method suitable for High Reflective Surface | |
US20120092463A1 (en) | Dual-frequency Phase Multiplexing (DFPM) and Period Coded Phase Measuring (PCPM) Pattern Strategies in 3-D Structured Light Systems, and Lookup Table (LUT) Based Data Processing | |
Hu et al. | Microscopic 3D measurement of shiny surfaces based on a multi-frequency phase-shifting scheme | |
CN105066906B (en) | A kind of quick high dynamic range method for three-dimensional measurement | |
CN106643562B (en) | Structural light stripes projective techniques based on time domain airspace hybrid coding | |
CN103292741A (en) | Structured light vision measurement method for 3D surface profiles of objects on the basis of K-means color clustering | |
US10757393B2 (en) | Method and apparatus for measuring characteristics of a surface topology | |
CN108596008B (en) | Face shake compensation method for three-dimensional face measurement | |
CN103940370B (en) | Target object 3 D information obtaining method based on cycle relatively prime hybrid coding | |
CN111536905B (en) | Monocular grating structure optical vision measurement method and system based on reference image | |
CN113379818A (en) | Phase analysis method based on multi-scale attention mechanism network | |
Lu et al. | Reflectance and shape from images using a collinear light source | |
Li et al. | An improved 2+ 1 phase-shifting algorithm | |
Yu et al. | Color phase order coding and interleaved phase unwrapping for three-dimensional shape measurement with few projected pattern | |
Zhou et al. | High-dynamic-range 3-D shape measurement with adaptive speckle projection through segmentation-based mapping | |
Fu et al. | High dynamic range structured light 3-D measurement based on region adaptive fringe brightness | |
Fernandez et al. | Absolute phase mapping for one-shot dense pattern projection | |
Fernandez et al. | A novel structured light method for one-shot dense reconstruction | |
Dizeu et al. | Frequency shift triangulation: a robust fringe projection technique for 3D shape acquisition in the presence of strong interreflections |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |