CN107907066B - Monocular six degree of freedom deformation monitoring system and method - Google Patents
Monocular six degree of freedom deformation monitoring system and method Download PDFInfo
- Publication number
- CN107907066B CN107907066B CN201711093842.6A CN201711093842A CN107907066B CN 107907066 B CN107907066 B CN 107907066B CN 201711093842 A CN201711093842 A CN 201711093842A CN 107907066 B CN107907066 B CN 107907066B
- Authority
- CN
- China
- Prior art keywords
- frame image
- pair
- label
- marks
- frame
- 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.)
- Active
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
- G01B11/16—Measuring arrangements characterised by the use of optical techniques for measuring the deformation in a solid, e.g. optical strain gauge
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of monocular six degree of freedom deformation monitoring system and methods, including target, vision collecting unit, central processing unit and output interface unit;Wherein: target is fixed on body surface to be monitored, which is provided with 5 pairs of circular marks pair, and all circular marks etc. are big;The surface of the target and the circular mark is two kinds of colors with gray difference, can extract circular mark from target image;5 pairs of circular marks are to the center, upside, downside, left and right side for being located at the target;Vision collecting unit is used to acquire the video stream data of target comprising ccd imaging sensor, diaphragm, optical lens group, the optical filter being arranged successively along optical path;Central processing unit is used to carry out the video stream data image procossing, and exports processing result by the output interface unit.Present invention combination monocular vision and close-range photogrammetry realize the on-line monitoring of object six degree of freedom deformation.
Description
Technical field
The invention belongs to close-shot photography measure technique fields, specifically, being related to a kind of monocular six degree of freedom deformation monitoring
System and method.
Background technique
Deformation refers to that object is generated the change in shape such as flexible, inclination, bending by external force.Deformation is widely present in the modern times and builds
Build and physical relief in, as bridge, dam, mine and skyscraper it is in bad repair with age caused by sink, the plastic deformations such as inclination,
Landform is because of landslide, rockfall, earthquake, subsidence etc. caused by natural calamity.Deformation may exist in a certain range,
Referred to as safe range.If deformation exceeds safe range, accident or disaster can be evolved into, to endanger national personal property
Safety.Therefore, being monitored on-line to the deformation situation of modern architecture and physical relief is just particularly important.
Close-range photogrammetry (Close Range Photogrammetry) technology refers to the mesh to object distance no more than 300 meters
Mark intake image, and utilize the information such as image processing techniques and size, the speed of the prior information of target measurement object.This hair
The bright main monocular six degree of freedom deformation monitoring technology for proposing complete set, by monocular vision and close-shot photography measure technique,
The multiple degrees of freedom during deformation monitoring is solved simultaneously, real-time, portability, blocks the technical problems such as sensitivity.
Summary of the invention
The object of the present invention is to provide a kind of monocular six degree of freedom deformation for combining monocular vision and close-range photogrammetry
Monitor system and method.
Monocular six degree of freedom deformation monitoring system provided by the invention, including target, vision collecting unit, central processing list
Member and output interface unit;Wherein:
Target is fixed on body surface to be monitored, which is provided with 5 pairs of circular marks pair, and all circular marks etc. are big;It is described
The surface of target and the circular mark is two kinds of colors with gray difference, can extract round mark from target image
Note;
5 pairs of circular marks are denoted as the center, upside, downside, left and right side that are located at the target respectively
Central marks to, it is upper label to, it is lower label to, it is left label to, it is right label pair;Wherein, central marks to, it is upper label to and subscript
Remember centering two circular marks distinguish horizontal alignment, and it is upper label to it is lower label to about central marks to symmetrical above and below point
Cloth;Left label is vertically aligned two circular marks with right label centering, and left label is marked to right to about center
Label is to bilateral symmetry;
Vision collecting unit is used to acquire the video stream data of target comprising passes along the CCD imaging that optical path is arranged successively
Sensor, diaphragm, optical lens group, optical filter;
Central processing unit is used to carry out the video stream data image procossing, and defeated by the output interface unit
Processing result out.
Monocular six degree of freedom deformation monitoring method provided by the invention, including to vision collecting unit described in claim 1
Video stream data collected is handled as follows:
S100 frame selects central marks pair, and the frame image of institute's frame choosing is denoted as initial frame image, and the rectangular area of institute's frame choosing is denoted as
Frame favored area extracts central marks to profile, based on extracted central marks to the priori position of profile and each label pair
Relationship calculates the target position information of initial frame image;The target position information is including 1. central marks to profile and the center of circle
The pixel distance of center circle and pixel radius of position, 2. central marks pair, 3. upper label are to, lower label to, left label to, right label pair
Center location;
Target position information of the S200 according to prior image frame in consecutive frame image, frame image after obtaining in each consecutive frame image
Target position information, to obtain the target position information of each frame image, this step further comprises:
S210 estimates central marks in rear frame image and is denoted as discreet area to the rectangular area at place, specifically:
When prior image frame is initial frame image in adjacent two field pictures, then the discreet area of rear frame image is prior image frame
Frame favored area;When prior image frame is not initial frame image in adjacent two field pictures, then according to the former frame of prior image frame to previous frame
In the time interval of image, central marks combine the discreet area of prior image frame to the velocity vector for making interframe displacement, estimate
The discreet area of frame image afterwards;
Calculating of the central marks to the velocity vector for making interframe displacement are as follows:
According to the center location of central marks pair in adjacent two field pictures, central marks are calculated to the displacement vector of interframe;
According to the displacement vector of the time interval of adjacent two field pictures and interframe, the velocity vector that central marks are displaced interframe is calculated;
S220 is split the discreet area of rear frame image, and extraction central marks are to profile, the target of frame image after calculating
Cursor position information;
It is relatively initial to obtain target in each frame image according to the target position information of each frame image and initial frame image by S300
The deformation quantity of frame image, the deformation quantity include the real standard displacement of central marks pair, practical vertical displacement amount, optical axis side
To actual displacement amount and target around horizontal direction, vertical direction, the rotation angle of optical axis direction.
Further, step S100 further comprises:
The artificial frame of S110 selects central marks pair, and the frame image of frame choosing is denoted as initial frame image, the rectangular area note of institute's frame choosing
For frame favored area;
S120 is split frame favored area, extracts central marks to profile and center location;
S130 calculates the pixel distance of center circle and pixel radius of central marks pair according to center location;
S140 according to each label pair priori position relationship, in initial frame image in search label to, lower label to, it is left
Label to it is right label pair, and obtain it is all label pair center location.
Further, in step S210, velocity vector that the central marks are displaced interframeAre as follows:
Wherein:
For the real standard displacement vector of central marks pair in adjacent two field pictures;
For the practical vertical displacement vectors of central marks pair in adjacent two field pictures;
For the practical optical axis direction displacement vector of central marks pair in adjacent two field pictures;
Fps indicates the inverse of the time interval of adjacent two field pictures.
Further, the real standard displacement of the central marks pair and practical vertical displacement amount use formulaIt calculates,
Wherein:
K is imaging scale;D is the practical distance of center circle of central marks pair, and D is prior information;DpIn in initial frame image
The pixel distance of center circle of centre label pair;
Δ x is real standard displacement of the current frame image relative to the central marks pair of initial frame image;
ΔxpPixel level displacement for current frame image relative to the central marks pair of initial frame image;
Δ y is practical vertical displacement amount of the current frame image relative to the central marks pair of initial frame image;
ΔypPixel vertical displacement amount for present frame relative to the central marks pair of initial frame image;
Further, rotation angle of the target around optical axis directionWherein, Δ ylWith Δ yrPoint
It left Wei not mark to the practical vertical displacement amount with right label pair, according to the target position confidence in each frame image and initial frame image
Breath obtains;HlrIt is left label to the distance with right label pair.
Further, rotation angle of the target around horizontal directionWherein, f is light
Learn the focal length of lens group;dt、db、d0In respectively each frame image upper label to, lower label to and central marks pair the pixel center of circle
Away from;htbFor it is upper label to it is lower label pair distance and central marks to the ratio between practical distance of center circle.
Further, rotation angle of the target around vertical directionWherein, f is light
Learn the focal length of lens group;dl、dr、d0Left label is to, right label to the pixel center of circle of, central marks pair in respectively each frame image
Away from;hldFor it is left label to it is right label pair distance and central marks to the ratio between practical distance of center circle.
Further, the method for the present invention further includes the steps that shadowing and abandons by blocking the frame image influenced;
The shadowing step includes:
To all circular marks in each frame image, calculate separately elemental area in its profile, pixel grey scale average value and
Profile axial ratio three elements, it is poor that the three elements value of each frame image is made with the three elements value of its prior image frame respectively, once have
The difference of a certain three elements value of circular mark is more than the corresponding preset threshold of the element, that is, being judged as to be blocked influences.
Further, the method for the present invention further includes being smoothed to the deformation quantity of n-th frame and later frame image
Step, specifically:
Deformation quantity of the current frame image with respect to initial frame image is calculated, and distinguishes phase with n-1 frame image before current frame image
Arithmetic mean of instantaneous value is asked to the deformation quantity of initial frame image, using arithmetic mean of instantaneous value as the sharpening result of current frame image;Wherein, n
According to camera parameter and processor performance in 50~100 ranges value.
Compared to the prior art, the invention has the advantages that and the utility model has the advantages that
(1) present invention realizes automation and portability on image taking and processing, six degree of freedom positioning and tracking, leads to
The prior information of monocular vision and target is crossed, while solving multiple degrees of freedom, real-time, portability, blocking the technical problems such as sensitivity,
Realize the on-line monitoring of six degree of freedom deformation.
(2) in terms of measurement accuracy, the measurement error absolute value that the present invention is displaced on X, Y direction is no more than
0.01mm;The measurement error absolute value of displacement in the Z-axis direction is no more than 0.1mm;Around X, the measurement error of Y-axis rotation angle is exhausted
0.1 ° is no more than to value;The measurement error absolute value of rotation angle is no more than 0.002 ° about the z axis.
(3) in terms of real-time, processing speed can reach 25 frame per second or more to the present invention on a general-purpose computer;It is being embedded in
It can reach 10 frame per second or more on formula platform.
(4) in terms of portability, the present invention is based on ARM microprocessor platforms, system can be fabricated to handheld device;
(5) in terms of blocking sensitivity, the present invention can whether there is according to the grayscale information real time discriminating of circular mark is blocked
Situation.
Detailed description of the invention
Fig. 1 is the concrete structure schematic diagram of present system;
Fig. 2 is the schematic diagram that target rotates about the z axis;
Fig. 3 is the schematic diagram that target is rotated around X-axis;
Fig. 4 is the schematic diagram that target is rotated around Y-axis;
Fig. 5 is image segmentation figure;
Fig. 6 is a kind of specific flow chart of the method for the present invention.
In figure, 1- target, 2- circular mark, 3-CCD imaging sensor, 4- diaphragm, 5- optical lens group, 6- optical filter,
7-USB cable, 8- central processing unit, 9-LVDS signaling interface, 10-HDMI interface.
Specific embodiment
In order to illustrate the embodiments of the present invention more clearly and/or technical solution in the prior art, attached drawing will be compareed below
Illustrate a specific embodiment of the invention.It should be evident that the accompanying drawings in the following description is only the embodiment of the present invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing, and obtain other embodiments.
It should be noted that monocular six degree of freedom deformation monitoring system of the present invention and method, wherein monocular refers to using only one
A camera, six degree of freedom then include displacement in dimensional Cartesian rectangular coordinate system in tri- axis direction of X, Y, Z and around X, Y, Z
The angle of three axis rotation.
See that Fig. 1, the monocular six degree of freedom deformation monitoring system include target 1, vision collecting unit, central processing unit
8 and output interface unit.The target 1, which is equipped with, waits big several circular marks 2, and the surface colour of target 1 and label 2 is gray scale
The two kinds of colors to differ greatly, in order to extract circular mark from target image.In present embodiment, the target
1 is a Thin Rectangular hardboard, and the surface of the target 1 and the circular mark 2 is respectively black and white.
More specifically, 1 plane of target is equipped with 5 pairs of circular marks pair.1st pair of white marking contraposition is in target 1
The heart is denoted as central marks pair, two circular mark horizontal alignment of central marks centering.2nd pair of white marking to and the 3rd dialogue colour code
Note to the upper side and lower side for being located at target 1, be denoted as respectively label to it is lower label pair, it is described it is upper label to it is described under
Label is to about the central marks, to symmetrical above and below, and the upper label is to two white markings point with the lower label centering
Other horizontal alignment.4th pair of white marking to and the 5th pair of white marking to the left and right side for being located at target 1, be denoted as respectively
Left label to right label pair, the left label to the right label to about the central marks to bilateral symmetry, and institute
Left label is stated to be vertically aligned two white markings with the right label centering.When monitoring starts, target 1 is fixed on
Body surface to be monitored, if object to be monitored generates deformation, the circular mark that will drive on target 1 is produced on six-freedom degree
Raw displacement and/or rotation.
The vision collecting unit is placed in the front of object to be monitored, for acquiring the video stream data of target 1.Vision
Acquisition unit includes the ccd imaging sensor 3 being arranged successively along optical path, diaphragm 4, optical lens group 5, optical filter 6.The optical filtering
Piece 6 is used to filter out Rayleigh scattering light and stray light.The optical lens group 5 should lay scene according to practical, i.e., for focusing
Measure the suitable focal length of selections and the sensitivity such as distance, light environment.The diaphragm 4 is for adjusting clear aperture, thus indirectly
Adjust image quality.The ccd imaging sensor 3 is used to convert two-dimensional optical image signal to dimensional video signal output.
In present embodiment, the ccd imaging sensor 3 use array image sensor, the optical filter 6 use bandwidth for
The narrow band filter of 15nm.
The vision collecting unit video stream data collected is transmitted to the central processing unit 8, this specific implementation
In mode, video stream data is transmitted to central processing unit 8 by USB cable 7, meanwhile, institute is given also by the USB cable 7
State ccd imaging sensor 3 and its drive circuitry.
The central processing unit 8 is used to carry out the video stream data image procossing, and passes through the output interface
Unit exports processing result.In present embodiment, the central processing unit 8 is the microprocessor based on ARM, is specifically adopted
With Tegra K1 processor, Linux+QT exploitation environment, cross compile Qt/E and OpenCV are built in central processing unit 8.
The output interface unit includes the LVDS signaling interface 9 and/or connection external display for connecting external liquid crystal touch screen
HDMI interface 10.
Fig. 6 show a kind of specific embodiment of monocular six degree of freedom deformation monitoring method of the present invention, including to described
Video stream data is handled as follows:
S100 frame selects central marks pair, and the frame image of institute's frame choosing is denoted as initial frame image, and the rectangular area of institute's frame choosing is denoted as
Frame favored area, obtains the location information of target in initial frame image, and the location information of the target includes 1. central marks pair
Profile and center location, 2. the pixel distance of center circle and pixel radius of central marks pair, 3. upper label are to, lower label to, left label
To, it is right label pair center location.
The pixel distance of center circle of the central marks pair refers to the pixel distance of center circle of two circular marks of central marks centering;It is described
The pixel radius of central marks pair refers to the pixel radius of two circular marks of central marks centering.
This step further comprises:
The artificial frame of S110 selects central marks pair, and the frame image of frame choosing is denoted as initial frame image, the rectangular area of institute's frame choosing
It is denoted as frame favored area.
S120 frame favored area is split with distinguish central marks to and background, and extract central marks pair profile and
Center location.
The realization of the watershed transform method based on label can be used in this step, specifically: become using the watershed based on label
Method divided frame favored area is changed, and fits circle contour, to obtain the profile and center location of central marks pair.
See Fig. 5, show the watershed transform schematic diagram based on label, wherein figure (a) is original frame favored area, figure
It (b) is schematic diagram after the regional partition of frame constituency.This method can be well according to the homogeneous region of label segmented image, and reference numerals are determined
Fixed finally obtained number of regions.After determining windowed regions, the area image is handled.Assuming that object in adjacent two frame time interval
Under the premise of displacement caused by body moves is less than the radius of circular mark, marks old frame center location for label a, then mark a must
In new frame circle;Marking the entire frame of rectangle frame is label b, then marks b that must not intersect with circle.After watershed transform, rectangle
It there will be only 2 homogeneous regions in frame, line of demarcation is circle contour, to divide obtained foreground image.Elliptic contour is fitted
Afterwards, it updates the center of circle and radius completes the tracking of new frame.
The pixel distance of center circle and pixel radius of S130 calculating central marks pair.
S140 according to it is upper label to, it is lower label to, it is left label to, it is right label to and central marks pair priori position close
System, in initial frame image search on label to, it is lower label to, it is left label to it is right label pair, obtain it is all label pair circle
Heart position.
Target position information of the S200 according to prior image frame in consecutive frame image, frame image after obtaining in consecutive frame image
Target position information.
In this step, according to the center location of central marks pair in adjacent two field pictures, central marks are calculated to interframe
Displacement calculates the direction and speed that central marks are displaced interframe based on the time interval of adjacent two field pictures;By the direction
Extraction with speed for target position information in next adjacent two field pictures.Rear frame image in the adjacent two field pictures is
Prior image frame in next adjacent two field pictures.
Based on above-mentioned, this step further comprises:
The discreet area of frame image after S210 is obtained in adjacent two field pictures, specifically:
If prior image frame is initial frame image in adjacent two field pictures, the discreet area of rear frame image is prior image frame
Frame favored area.If prior image frame is not initial frame image in adjacent two field pictures, according to the former frame of prior image frame to previous frame
In the time interval of image, central marks combine the discreet area of prior image frame to the velocity vector for making interframe displacement, estimate
The discreet area of frame image afterwards.
Calculating of the central marks to the velocity vector for making interframe displacement are as follows:
According to the center location of central marks pair in adjacent two field pictures, central marks are calculated to the displacement vector of interframe;
According to the displacement vector of the time interval of adjacent two field pictures and interframe, the velocity vector that central marks are displaced interframe is calculated.
The velocity vector that central marks are displaced interframeCalculating are as follows:
In formula (1):
For the real standard displacement vector of central marks pair in adjacent two field pictures;
For the practical vertical displacement vectors of central marks pair in adjacent two field pictures;
For the practical optical axis direction displacement vector of central marks pair in adjacent two field pictures;
Fps indicates the inverse of the time interval of adjacent two field pictures.
The deformation of object to be monitored has certain directionality and continuity, therefore, can by analyzing adjacent two field pictures,
The velocity vector that central marks are displaced interframe, the i.e. velocity vector of target interframe displacement are calculated, to predict deformation tendency.Root
The general area in next frame image where target can be estimated according to deformation tendency, and the search center that individually opens a window to the general area
Label is to profile.In the higher situation of image resolution ratio, windowing search can reduce calculation amount, improve search efficiency.
S220 is split the discreet area of rear frame image, extracts the profile and center location of central marks pair, and counts
Calculate the location information of target.The watershed transform method based on label specifically can be used and realize segmentation.
S300 calculates each frame image with respect to initial frame image according to the target position information of each frame image and initial frame image
Deformation quantity, the deformation quantity includes that (Δ x, Δ y, Δ z) and 3 rotation angles (α, beta, gamma) are, it can be achieved that space 6 for 3 displacements
The resolving of a freedom degree complexity posture.
The Computing Principle of deformation quantity will be illustrated below.
(1) calculating of the real standard displacement Δ x of central marks pair and practical vertical displacement amount Δ y.
When target moves in perpendicular to optical axial plane, the entirety of the displacement characterization target of central marks pair can be used
Displacement.According to geometric similarity method, because of the imaging scale of the displacement of central marks pair and the pixel distance of center circle of central marks pair
It is identical, so, the displacement of target planar can be calculated according to prior information of the central marks to pixel distance of center circle, as follows:
In formula (2):
K is imaging scale;D is the practical distance of center circle of central marks pair, and D is prior information;DpIn in initial frame image
The pixel distance of center circle of centre label pair;
Δ x is real standard displacement of the current frame image relative to the central marks pair of initial frame image;ΔxpTo work as
Pixel level displacement of the prior image frame relative to the central marks pair of initial frame image;
Δ y is practical vertical displacement amount of the current frame image relative to the central marks pair of initial frame image;ΔypTo work as
Pixel vertical displacement amount of the previous frame frame image relative to the central marks pair of initial frame image.
Here, the pixel level displacement of central marks pair and pixel vertical displacement amount refer to the circle center line connecting of central marks pair
The horizontal displacement and pixel vertical displacement amount at midpoint.
(2) calculating of the actual displacement amount Δ z of the optical axis direction of central marks pair.
Have by tracking central marks object distance u practical to calculating in conjunction with Gauss image-forming principle and triangle correspondence theorem:
In formula (3):
U is practical object distance, i.e., the distance of object to be monitored to optical lens group optical center;
F is the focal length of optical lens group;
D is the practical distance of center circle of central marks pair in current frame image;
D is central marks in current frame image to the imaging size of distance of center circle;
M is pixel dimension, is the constant parameter of ccd sensor;
N is the pixel number of imaging.
Central marks are calculated to the actual displacement amount Δ z in Z-direction using object distance, i.e., target is in Z in current frame image
The actual displacement amount of axis direction:
Δ z=u-u0 (4)
In formula (4): u0For the object distance of initial frame image;U is the object distance of current frame image.
(3) calculating of the rotation angle γ of target about the z axis.
See Fig. 2, show the schematic diagram that target rotates about the z axis.Wherein, Z axis and optical axis coincidence, rotation is being hung down about the z axis
Directly in the plane of optical axis around central marks to rotation.ΔylIt is denoted as the practical vertical displacement amount of left label pair, Δ yrIt is denoted as right mark
The practical vertical displacement amount of note pair, vertical displacement amount refers both to displacement of the current frame image with respect to initial frame image here.Zuo Biao
Remember to the distance with right label pair it is known that left label is 5D to the distance with right label pair in present embodiment.So
The rotation angle γ of target about the z axis are as follows:
(4) calculating of the target around the rotation angle α of X-axis.
See Fig. 3, show the schematic diagram that target is rotated around X-axis.When target is rotated around X-axis, can according to it is upper label to
The pixel distance of center circle of lower label pair calculates target around the rotation angle of X-axis.ΔztWith Δ zbBe denoted as respectively label to it is lower label pair
Actual displacement amount on Z axis, it is upper label to it is lower label pair between distance it is known that be 5D, acquire rotation angle α of the target around X-axis
Are as follows:
In formula (6), ut、ub、u0Respectively upper label to, lower label to and central marks pair practical object distance;dt、db、d0
Respectively in current frame image upper label to, lower label to and central marks pair pixel distance of center circle.
(5) calculating of the target around the rotation angle β of Y-axis.
See Fig. 4, show the schematic diagram that target is rotated around Y-axis.Target is rotary around Y-axis, similarly, can be according to left label
Target is calculated around the rotation angle of Y-axis to the pixel distance of center circle with right label pair.Rotation angle β of the target around Y-axis can be acquired are as follows:
In formula (7), Δ zlWith Δ zrRespectively left label is marked to right to the displacement on Z axis;ulAnd urRespectively
Left label is to the practical object distance with right label pair;dlAnd drLeft label is to the pixel with right label pair respectively in current frame image
Distance of center circle.
Good sighting condition is generally required by the system of measurement means of monocular vision, the presence of shelter can be to measurement
Effect generates certain image.If not blocking method of discrimination accordingly, or even the measurement result of mistake can be generated.Therefore,
The present invention also provides a kind of schemes, including shadowing and abandon by the frame image influenced is blocked, to avoid mistake is generated
As a result generation.This method specifically: to all circular marks in each frame image, calculate separately elemental area in its profile,
Pixel grey scale average value and profile axial ratio three elements, the three elements value of each frame image is wanted with the three of its prior image frame respectively
It is poor that element value is made, once the difference for having a certain three elements value of circular mark is more than the corresponding preset threshold of the element, then judges quilt
Shelter then abandons the frame image.
For the stability for improving monitoring result, measurement error caused by avoiding because of air shake, thermal noise etc., the present invention is also
A kind of preferred embodiment is proposed, including the smoothing step to n-th frame and later frame image deformation amount.Specifically: it calculates
Present image with respect to initial frame image deformation quantity, and with n-1 frame image before current frame image respectively with respect to initial frame image
Deformation quantity seeks arithmetic mean of instantaneous value, using arithmetic mean of instantaneous value as the sharpening result of current frame image;N according to processor performance 50~
Value in 100 ranges.
The monocular six degree of freedom that a kind of software and hardware being made of light, mechanical, electrical, calculation equipment integration that the present invention designs has both
Deformation monitoring system provides a kind of novel online prison for the deformation of the modern architectures such as bridge, dam, side slope and physical relief
Survey mode.It devises the target comprising 5 pairs of circular marks pair, and the displacement of spatial triaxial is calculated by tracking circular mark
And rotation angle.In terms of data processing, speed and the direction of interframe movement are made by target to estimate its next frame central marks
To the region at place;The profile of circular mark is extracted in windowed regions by watershed algorithm, and carries out blocking differentiation, avoids producing
Raw erroneous measurements.This method realizes automation and portability on image taking and processing, target positioning and tracking, leads to
The prior information of monocular vision and target is crossed, while solving multiple degrees of freedom, real-time, portability, blocking the technical problems such as sensitivity.
Claims (10)
1. monocular six degree of freedom deformation monitoring system, it is characterized in that:
Including target, vision collecting unit, central processing unit and output interface unit;Wherein:
Target is fixed on body surface to be monitored, which is provided with 5 pairs of circular marks pair, and all circular marks etc. are big;The target
Surface with the circular mark is two kinds of colors with gray difference, can extract circular mark from target image;
5 pairs of circular marks are denoted as center to the center, upside, downside, left and right side that are located at the target respectively
Label to, it is upper label to, it is lower label to, it is left label to, it is right label pair;Wherein, central marks to, it is upper label to it is lower label pair
In two circular marks distinguish horizontal alignment, and it is upper label to it is lower label to about central marks to distribution symmetrical above and below;It is left
It marks and two circular marks with right label centering is vertically aligned, and left label is marked to right to about central marks pair
Symmetrically;
Vision collecting unit is used to acquire the video stream data of target comprising the ccd imaging sensor that is arranged successively along optical path,
Diaphragm, optical lens group, optical filter;
Central processing unit be used to the video stream data carry out image procossing, and by the output interface unit output at
Manage result.
2. using the monocular six degree of freedom deformation monitoring method of monocular six degree of freedom deformation monitoring system described in claim 1,
It is characterized in:
Including vision collecting unit video stream data collected described in claim 1 is handled as follows:
S100 frame selects central marks pair, and the frame image of institute's frame choosing is denoted as initial frame image, and the rectangular area of institute's frame choosing is denoted as frame choosing
Region is extracted central marks to profile, is closed based on priori position of the extracted central marks to profile and each label pair
System calculates the target position information of initial frame image;The target position information is including 1. central marks to profile and center of circle position
Set, 2. the pixel distance of center circle and pixel radius of central marks pair, 3. upper label to, lower label to, left label to, right label pair
Center location;
Target position information of the S200 according to prior image frame in consecutive frame image, the target of frame image after obtaining in each consecutive frame image
Cursor position information, to obtain the target position information of each frame image, this step further comprises:
S210 estimates central marks in rear frame image and is denoted as discreet area to the rectangular area at place, specifically:
When prior image frame is initial frame image in adjacent two field pictures, then the discreet area of rear frame image is that the frame of prior image frame selects
Region;When prior image frame is not initial frame image in adjacent two field pictures, then according to the former frame of prior image frame to prior image frame
Time interval in, central marks to make interframe displacement velocity vector, and combine prior image frame discreet area, estimate rear frame
The discreet area of image;
Calculating of the central marks to the velocity vector for making interframe displacement are as follows:
According to the center location of central marks pair in adjacent two field pictures, central marks are calculated to the displacement vector of interframe;According to
The time interval of adjacent two field pictures and the displacement vector of interframe calculate the velocity vector that central marks are displaced interframe;
S220 is split the discreet area of rear frame image, and extraction central marks are to profile, the target position of frame image after calculating
Confidence breath;
S300 obtains in each frame image target with respect to initial frame figure according to the target position information of each frame image and initial frame image
The deformation quantity of picture, the deformation quantity include the real standard displacement of central marks pair, practical vertical displacement amount, optical axis direction
Actual displacement amount and target are around horizontal direction, vertical direction, the rotation angle of optical axis direction.
3. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
Step S100 further comprises:
The artificial frame of S110 selects central marks pair, and the frame image of frame choosing is denoted as initial frame image, and the rectangular area of institute's frame choosing is denoted as frame
Favored area;
S120 is split frame favored area, extracts central marks to profile and center location;
S130 calculates the pixel distance of center circle and pixel radius of central marks pair according to center location;
S140 according to each label pair priori position relationship, in initial frame image in search label to, lower label to, left label
To with it is right label pair, and obtain it is all label pair center location.
4. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
In step S210, velocity vector that the central marks are displaced interframeAre as follows:
Wherein:
For the real standard displacement vector of central marks pair in adjacent two field pictures;
For the practical vertical displacement vectors of central marks pair in adjacent two field pictures;
For the practical optical axis direction displacement vector of central marks pair in adjacent two field pictures;
Fps indicates the inverse of the time interval of adjacent two field pictures.
5. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
The real standard displacement of the central marks pair and practical vertical displacement amount use formulaIt calculates,
Wherein:
K is imaging scale;D is the practical distance of center circle of central marks pair, and D is prior information;DpFor central marks in initial frame image
Pair pixel distance of center circle;
Δ x is real standard displacement of the current frame image relative to the central marks pair of initial frame image;
ΔxpPixel level displacement for current frame image relative to the central marks pair of initial frame image;
Δ y is practical vertical displacement amount of the current frame image relative to the central marks pair of initial frame image;
ΔypPixel vertical displacement amount for present frame relative to the central marks pair of initial frame image.
6. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
Rotation angle of the target around optical axis directionWherein, Δ ylWith Δ yrRespectively it is left label to
The practical vertical displacement amount of right label pair, according to the target position information acquisition in each frame image and initial frame image;HlrFor a left side
It marks to the distance with right label pair.
7. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
Rotation angle of the target around horizontal directionWherein, f is the focal length of optical lens group;
dt、db、d0In respectively each frame image upper label to, lower label to and central marks pair pixel distance of center circle;htbFor upper label pair
Distance and central marks with lower label pair is to the ratio between practical distance of center circle.
8. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
Rotation angle of the target around vertical directionWherein, f is the focal length of optical lens group;
dl、dr、d0Left label is to, right label to the pixel distance of center circle of, central marks pair in respectively each frame image;hldFor left label pair
Distance and central marks with right label pair is to the ratio between practical distance of center circle.
9. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
Further include the steps that shadowing and abandons by blocking the frame image influenced;
The shadowing step includes:
To all circular marks in each frame image, elemental area, pixel grey scale average value and the profile in its profile are calculated separately
Axial ratio three elements, it is poor that the three elements value of each frame image is made with the three elements value of its prior image frame respectively, once there is circle
The difference of a certain three elements value of label is more than the corresponding preset threshold of the element, that is, being judged as to be blocked influences.
10. monocular six degree of freedom deformation monitoring method as claimed in claim 2, it is characterized in that:
Further include the steps that the deformation quantity to n-th frame and later frame image is smoothed, specifically:
Deformation quantity of the current frame image with respect to initial frame image is calculated, and relatively first respectively with n-1 frame image before current frame image
The beginning deformation quantity of frame image seeks arithmetic mean of instantaneous value, using arithmetic mean of instantaneous value as the sharpening result of current frame image;Wherein, n according to
Camera parameter and the processor performance value in 50~100 ranges.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711093842.6A CN107907066B (en) | 2017-11-08 | 2017-11-08 | Monocular six degree of freedom deformation monitoring system and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711093842.6A CN107907066B (en) | 2017-11-08 | 2017-11-08 | Monocular six degree of freedom deformation monitoring system and method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107907066A CN107907066A (en) | 2018-04-13 |
CN107907066B true CN107907066B (en) | 2019-08-02 |
Family
ID=61843766
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711093842.6A Active CN107907066B (en) | 2017-11-08 | 2017-11-08 | Monocular six degree of freedom deformation monitoring system and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107907066B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108927807B (en) * | 2018-08-14 | 2020-08-07 | 河南工程学院 | Robot vision control method based on point characteristics |
CN109084698A (en) * | 2018-10-25 | 2018-12-25 | 北京测科空间信息技术有限公司 | A kind of method and system for engineering works deformation monitoring |
CN109737883A (en) * | 2018-12-21 | 2019-05-10 | 成都蕴才汇智科技有限责任公司 | A kind of three-dimensional deformation dynamic measurement system and measurement method based on image recognition |
CN109655003A (en) * | 2019-01-25 | 2019-04-19 | 成都大亦科技有限公司 | A kind of multidimensional image identification displacement measuring device and method based on calibration substrate |
CN110332887B (en) * | 2019-06-27 | 2020-12-08 | 中国地质大学(武汉) | Monocular vision pose measurement system and method based on characteristic cursor points |
CN114659482B (en) * | 2020-12-22 | 2023-06-13 | 核动力运行研究所 | High-precision circular outline dimension measurement algorithm based on non-uniform discrete data |
CN113407757B (en) * | 2021-06-23 | 2022-07-22 | 重庆世纪科怡科技股份有限公司 | Image retrieval method and device based on computer |
CN113344924B (en) * | 2021-08-05 | 2021-10-29 | 金成技术有限公司 | Structural member production control method and system based on image processing |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2835603A1 (en) * | 2002-02-01 | 2003-08-08 | Holo 3 | Optical measurement of displacement and or deformation of an object, e.g. for civil engineering and industrial use, whereby a simple camera is used in conjunction with a calibrated target fixed to the object to be monitored |
CN201463843U (en) * | 2009-07-10 | 2010-05-12 | 上海同岩土木工程科技有限公司 | Digital photography-based geotechnical engineering deformation remote monitoring system |
CN101832760A (en) * | 2010-04-23 | 2010-09-15 | 清华大学 | Remote three-dimensional micro-deformation visual on-line monitoring method and system |
CN202928571U (en) * | 2012-11-05 | 2013-05-08 | 上海米度测控科技有限公司 | Photographic measurement system for long-distance micro-deformation monitoring |
CN105091772A (en) * | 2015-05-26 | 2015-11-25 | 广东工业大学 | Plane object two-dimension deflection measuring method |
-
2017
- 2017-11-08 CN CN201711093842.6A patent/CN107907066B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2835603A1 (en) * | 2002-02-01 | 2003-08-08 | Holo 3 | Optical measurement of displacement and or deformation of an object, e.g. for civil engineering and industrial use, whereby a simple camera is used in conjunction with a calibrated target fixed to the object to be monitored |
CN201463843U (en) * | 2009-07-10 | 2010-05-12 | 上海同岩土木工程科技有限公司 | Digital photography-based geotechnical engineering deformation remote monitoring system |
CN101832760A (en) * | 2010-04-23 | 2010-09-15 | 清华大学 | Remote three-dimensional micro-deformation visual on-line monitoring method and system |
CN202928571U (en) * | 2012-11-05 | 2013-05-08 | 上海米度测控科技有限公司 | Photographic measurement system for long-distance micro-deformation monitoring |
CN105091772A (en) * | 2015-05-26 | 2015-11-25 | 广东工业大学 | Plane object two-dimension deflection measuring method |
Non-Patent Citations (2)
Title |
---|
基于机器视觉技术的振动台试验变形位移测量方法研究;陈若珠等;《计算机测量与控制》;20161231(第6期);第56-59页 |
基于机器视觉的桥梁形变在线监测技术研究;王翔等;《公路工程》;20140228;第39卷(第1期);第198-201页 |
Also Published As
Publication number | Publication date |
---|---|
CN107907066A (en) | 2018-04-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107907066B (en) | Monocular six degree of freedom deformation monitoring system and method | |
CN106441138B (en) | The deformation monitoring method of view-based access control model measurement | |
CA3071299C (en) | Initial alignment system and method for strap-down inertial navigation of shearer based on optical flow method | |
CN105225482B (en) | Vehicle detecting system and method based on binocular stereo vision | |
CN105447853A (en) | Flight device, flight control system and flight control method | |
CN105698699A (en) | A binocular visual sense measurement method based on time rotating shaft constraint | |
CN106643699A (en) | Space positioning device and positioning method in VR (virtual reality) system | |
CN114018236B (en) | Laser vision strong coupling SLAM method based on self-adaptive factor graph | |
CN105346706A (en) | Flight device, and flight control system and method | |
CN104964708B (en) | A kind of road surface pit detection method based on vehicle-mounted binocular vision | |
CN102927917A (en) | Multi-view vision measurement method of iron tower | |
CN110146030A (en) | Side slope surface DEFORMATION MONITORING SYSTEM and method based on gridiron pattern notation | |
JP5079547B2 (en) | Camera calibration apparatus and camera calibration method | |
CN109212377A (en) | A kind of high-tension line obstacle recognition method, device, crusing robot | |
Ruotsalainen et al. | Heading change detection for indoor navigation with a smartphone camera | |
CN110851978B (en) | Camera position optimization method based on visibility | |
CN111598097A (en) | Instrument position and reading identification method and system based on robot vision | |
CN107103582B (en) | The matching process of robot visual guidance positioning image characteristic point | |
CN112017229B (en) | Camera relative pose solving method | |
CN115937243A (en) | Downstream river image deformation monitoring method based on point cloud three-dimensional data | |
CN114569114A (en) | Height measuring method and device | |
CN115731343A (en) | Multi-person multi-view 3D reconstruction method based on top view image segmentation | |
CN114627398A (en) | Unmanned aerial vehicle positioning method and system based on screen optical communication | |
RU2685761C1 (en) | Photogrammetric method of measuring distances by rotating digital camera | |
CN115393387A (en) | Building displacement monitoring method and device |
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 |