CN109919856A - Bituminous pavement construction depth detection method based on binocular vision - Google Patents
Bituminous pavement construction depth detection method based on binocular vision Download PDFInfo
- Publication number
- CN109919856A CN109919856A CN201910053244.9A CN201910053244A CN109919856A CN 109919856 A CN109919856 A CN 109919856A CN 201910053244 A CN201910053244 A CN 201910053244A CN 109919856 A CN109919856 A CN 109919856A
- Authority
- CN
- China
- Prior art keywords
- gray level
- pixel
- correction
- level image
- image
- 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
- 238000010276 construction Methods 0.000 title claims abstract description 38
- 238000001514 detection method Methods 0.000 title claims abstract description 30
- 238000012937 correction Methods 0.000 claims abstract description 75
- 241000208340 Araliaceae Species 0.000 claims abstract description 13
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims abstract description 13
- 235000003140 Panax quinquefolius Nutrition 0.000 claims abstract description 13
- 235000008434 ginseng Nutrition 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 50
- 239000013598 vector Substances 0.000 claims description 9
- 238000013519 translation Methods 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 230000003287 optical effect Effects 0.000 claims description 4
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000003786 synthesis reaction Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000002159 abnormal effect Effects 0.000 claims 1
- 238000012360 testing method Methods 0.000 abstract description 8
- 238000000034 method Methods 0.000 description 29
- 238000005516 engineering process Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000004576 sand Substances 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 241000226585 Antennaria plantaginifolia Species 0.000 description 1
- 239000010426 asphalt Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003760 hair shine Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000011900 installation process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000011148 porous material Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G06T5/70—
-
- G06T5/80—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
Abstract
The present invention provides a kind of bituminous pavement construction depth detection method based on binocular vision, include the following steps: 100. acquisitions or so, two video cameras internal reference and outer ginseng；200. acquire the left color image and right color image of bituminous pavement using two video cameras in left and right respectively；Left color image and right color image are processed into left gray level image and right gray level image by 300. respectively；400. carry out distortion correction to left gray level image and right gray level image respectively, obtain the left gray level image of the first correction and the first right gray level image of correction；500. carry out three-dimensional correction to the left gray level image of the first correction and the first right gray level image of correction respectively, obtain the left gray level image of the second correction and the second right gray level image of correction；The left gray level image of 600. pair of second correction and the second right gray level image of correction carry out Stereo matching；700. eliminate the value of Stereo matching mistake；800. amendment shooting angle errors；900. calculate the construction depth of bituminous pavement.The present invention have many advantages, such as rapidly and efficiently, be not easy to be disturbed, price economy and testing result it is more accurate.
Description
Technical field
The present invention relates to the detection techniques of bituminous pavement construction depth in road project, more particularly, to a kind of base
In the bituminous pavement construction depth detection method of binocular vision.
Background technique
The antiskid performance of bituminous pavement influences significantly traffic safety, and construction depth is evaluation Bitumen Pavement Anti-Skid Performance
Important indicator.The construction depth of bituminous pavement refers to the mean depth of the rough open pores of road surface, reflects road
The degree of roughness in face.The too small antiskid performance that will reduce bituminous pavement of pavement structural depth, not only results in automobile and beats
Sliding phenomenon, and will increase the braking distance of automobile, seriously affect traffic safety.
Currently, the detection method of bituminous pavement construction depth mainly sands method, laser texture meter method and digitized map
As three kinds of method.It is simple to sand method principle, measurement is convenient, but extremely time-consuming；Although laser texture meter method precision is higher,
It is to need special equipment, it is expensive；Digital image method detects rapidly and efficiently, but is susceptible to ambient light and shines and road surface itself
The interference of color.Obviously, there is take a long time or expensive and easy existing bituminous pavement construction depth detection method
The problems such as being disturbed.
Therefore, study it is a kind of rapidly and efficiently, be not easy to be disturbed, the pitch construction depth detection method of price economy has must
The property wanted.
Summary of the invention
In order to overcome the shortcomings of the prior art, the present invention provide it is a kind of rapidly and efficiently, be not easy to be disturbed and price warp
The bituminous pavement construction depth detection method of Ji.
The technical scheme is that providing a kind of bituminous pavement construction depth detection method based on binocular vision, wrap
Include following steps:
100. internal reference and the outer ginseng of acquisition left and right two video camera；
200. acquire the left color image and right color image of bituminous pavement using two video cameras in left and right respectively；
Left color image and right color image are processed into left gray level image and right gray level image by 300. respectively；
400., according to the internal reference for controlling two video cameras, carry out distortion school to left gray level image and right gray level image respectively
Just, the left gray level image of the first correction and the first right gray level image of correction are obtained；
500. internal references according to the video camera of left and right two and outer ginseng, respectively to the left gray level image of the first correction and the first correction
Right gray level image carries out three-dimensional correction, obtains the left gray level image of the second correction and the second right gray level image of correction；
The left gray level image of 600. pair of second correction and the second right gray level image of correction carry out Stereo matching, identify the second school
Corresponding pixel on just left gray level image and the second right gray level image of correction, calculates parallax value d, is calculated and schemed according to parallax value d
As upper each pixel under camera coordinate system apart from the height value of camera plane, generate one and include each picture on image
The model matrix M of the pixel coordinate of vegetarian refreshments and its corresponding height value information, recovers the threedimensional model on road surface；
A threshold value is arranged in the difference coefficient value of two neighboring pixel height value in 700. pairs of model matrixs, determines three-dimensional
Position with error value is modified using value of the median filtering window to three-dimensional matching error, eliminates Stereo matching mistake
Value；
800. couples of model matrix M carry out plane fitting, and model matrix M and fit Plane are subtracted each other, when amendment acquires image
Due to camera optical axis and road surface out of plumb bring shooting angle error；
900. calculate the construction depth of bituminous pavement.
Further include following steps in above-mentioned steps 300 as improvement of the present invention:
301. are pressed left color image and right color image by red (R), green (G), blue three channels (B) respectively
After formula calculates, the left single channel gray level image and right single channel gray level image that are converted to；
F (x, y)=R (x, y) × 0.299+G (x, y) × 0.587+B (x, y) × 0.114；
Wherein, f (x, y) is the gray value of pixel, and R (x, y), G (x, y), B (x, y) are respectively the red, green of pixel
Color, blue three channels value.
Further include following steps in above-mentioned steps 300 as improvement of the present invention:
302. carry out denoising to left single channel gray level image and right single channel gray level image using median filtering, obtain
Left gray level image and right gray level image.
Further include following steps in above-mentioned steps 400 as improvement of the present invention:
401. determine distortion factor k according to the internal reference of video camera according to the following formula1、k2:
Wherein, k1、k2For the distortion factor of video camera, u, v are distortionless pixel coordinate, and x, y are distortionless continuous image
Plain coordinate, u0、v0For the pixel coordinate of video camera principal point,For the pixel coordinate after distortion；
402. utilize resulting distortion of camera coefficient k1、k2As the following formula, respectively to left gray level image and right gray level image into
Line distortion correction:
Further include following steps in above-mentioned steps 500 as improvement of the present invention:
501. determine the relative positional relationship between two video cameras in left and right:
502. are converted using Douglas Rodríguez, and relative rotation matrices are decomposed into left image and the respective synthesis of right image
Spin matrix rl、rr；
503. calculate the respective spin matrix R of left and right two imageslt、Rrt, by left image according to spin matrix RlRotation, will
Right image is according to spin matrix RrIt is rotated, keeps the polar curve of two images horizontal, and pole is at infinity, complete three-dimensional school
Just.
Further include following steps in above-mentioned steps 600 as improvement of the present invention:
After 601. using each pixel on half global registration algorithm (SGBM) traversal image, the second correction left side is identified
The same pixel on gray level image and the second right gray level image of correction, and calculate the parallax value d of pixel；
Wherein, xl、xrThe respectively same pixel is on the left gray level image of the second correction and the second right gray level image of correction
Pixel horizontal axis coordinate, zcFor scale factor；
602. calculate height value z of each pixel apart from camera plane with following formula；
Wherein, TxFor component of the relative translation vector T in X direction, unit mm indicates two video cameras in left and right
Between horizontal distance, f is focal length of camera, and unit mm, d are parallax value, unit mm；
The height value of all pixels point is formed a model matrix M by 603., recovers the threedimensional model on road surface, wherein mould
Type matrix M is one and contains the matrix of the pixel coordinate of each pixel and its corresponding pixel height value on image.
Further include following steps in above-mentioned steps 700 as improvement of the present invention:
701. are calculated as follows the single order difference coefficient of the corresponding height value of pixel, determine the position of Stereo matching erroneous pixel point
It sets；
K=(zi+1-zi)/(xi+1-xi)；
Wherein, k is the single order difference coefficient of pixel, xi, xi+1For i-th and the pixel abscissa of i+1 pixel, zi, zi+1
For i-th and the height value of i+1 pixel, unit mm.
702. point by single order difference coefficient greater than 1 is defined as the point of matching error, using 7 × 7 filter window, matching is wrong
Point accidentally is placed on window center, and the height value of all pixels point in window is arranged from small to large, calculates picture in window
The intermediate value of vegetarian refreshments height value exports replaced pixel height value with the value of intermediate value replacement matching error.
Further include following steps in above-mentioned steps 800 as improvement of the present invention:
The pixel height value of 801. couples of model matrix M does plane fitting, and the parameter a of fit Plane is calculated as follows out1,
a2, a3；
Wherein, xi, yiFor the pixel coordinate of ith pixel point, ziFor the height value of ith pixel point, n is total in matrix
Pixel number.
Further include following steps in above-mentioned steps 800 as improvement of the present invention:
802. calculate the height value of revised each pixel using following formula, complete shooting angle error correction；
hi=zi-a1xi-a2yi-a3；
Wherein, zi、hiHeight value respectively before shooting angle amendment with revised ith pixel point, unit mm.
As improvement of the present invention, in above-mentioned steps 900, the construction depth for calculating bituminous pavement is carried out as the following formula
Hp, unit mm；
Wherein, hmaxFor the maximum value of pixel height value, unit mm, hiFor the height value of ith pixel point, unit
The line number and columns for being model matrix M for mm, m and n.
The present invention acquires a left side for bituminous pavement by two video cameras in left and right due to using two video cameras in left and right respectively
Color image and right color image, successively carry out gray proces, distortion correction, three-dimensional correction, the threedimensional model for recovering road surface,
It eliminates the value of Stereo matching mistake, correct the shooting angle error of video camera, finally calculate the construction depth of bituminous pavement；It is examining
It is illuminated by the light smaller with the influence of road surface self color during survey, is not only capable of measuring bituminous pavement construction depth, while can also
The threedimensional model on road surface is recovered, can more intuitively reflect the technology status information of bituminous pavement, is referred to for testing staff；
It overcomes laser texture meter method and needs disadvantage using special equipment and expensive, utilize a pair of common video camera mirror
Detection can be completed in head；Overcome traditional-handwork sand method and it is electronic sand method detection speed it is relatively slow and by artificial subjective impact compared with
Big disadvantage, have many advantages, such as rapidly and efficiently, be not easy to be disturbed, price economy and testing result it is more accurate.
Detailed description of the invention
Fig. 1 is process blocks schematic diagram of the invention.
Fig. 2 is that the present invention demarcates gridiron pattern used.
Planar structure schematic diagram when Fig. 3 is two camera operations in left and right in the present invention.
Fig. 4 is the left gray level image and right gray level image in the present invention after gray proces.
Fig. 5 is the left gray level image of the second correction and the second correction right side after completing distortion correction and three-dimensional correction in the present invention
Gray level image.
Fig. 6 is the tested region road surface model figure in the present invention comprising Stereo matching error value.
Fig. 7 is the tested region road surface model figure after the value for eliminating Stereo matching mistake in the present invention.
Fig. 8 is that the tested region road surface model figure after video camera shooting angle error is corrected in the present invention.
Specific embodiment
In the description of the present invention, it is to be understood that, "center" in term, "upper", "lower", "front", "rear", " left side ",
The orientation or positional relationship of instructions such as " right sides " is to be based on the orientation or positional relationship shown in the drawings, and is merely for convenience of describing this hair
Bright and simplified description, rather than the device or component of indication or suggestion meaning must have a particular orientation, with specific orientation
Construction and operation, therefore be not considered as limiting the invention.In addition, term " first ", " second " are used for description purposes only,
It is not understood to indicate or imply relative importance.
In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, term " installation " " connects
Connect ", " connected " shall be understood in a broad sense, for example, it may be being fixedly connected, be also possible to dismantling connection, or be integrally connected；It can be with
It is mechanical connection, is also possible to be electrically connected；It can be directly connected, be also possible to indirectly connected through an intermediary, can be
The connection of two component internals.For the ordinary skill in the art, above-mentioned term can be understood at this with concrete condition
The concrete meaning of invention.
Referring to Figure 1, Fig. 1 it is revealed be a kind of bituminous pavement construction depth detection method based on binocular vision stream
Cheng Tu, the bituminous pavement construction depth detection method based on binocular vision include the following steps:
100. internal reference and the outer ginseng of acquisition left and right two video camera；
200. acquire the left color image and right color image of bituminous pavement using two video cameras in left and right respectively；
Left color image and right color image are processed into left gray level image and right gray level image (refers to figure by 300. respectively
4)；
400., according to the internal reference for controlling two video cameras, carry out distortion school to left gray level image and right gray level image respectively
Just, the left gray level image of the first correction and the first right gray level image of correction are obtained；
500. internal references according to the video camera of left and right two and outer ginseng, respectively to the left gray level image of the first correction and the first correction
Right gray level image carries out three-dimensional correction, obtains the left gray level image of the second correction and the second right gray level image (referring to Fig. 5) of correction；
The left gray level image of 600. pair of second correction and the second right gray level image of correction carry out Stereo matching (referring to Fig. 6),
It identifies corresponding pixel on the left gray level image of the second correction and the second right gray level image of correction, calculates the parallax value of pixel
D is generated under camera coordinate system apart from the height value of camera plane according to each pixel on parallax value d calculating image
One includes the pixel coordinate of each pixel and its model matrix M of corresponding height value information on image, recovers road surface
Threedimensional model；
A threshold value is arranged in the difference coefficient value of two neighboring pixel height value in 700. pairs of model matrixs, determines three-dimensional
Position with erroneous point is modified using value of the median filtering window to three-dimensional matching error, eliminates Stereo matching mistake
Value (refers to Fig. 7)；
800. couples of model matrix M carry out plane fitting, and model matrix M and fit Plane are subtracted each other, when amendment acquires image
Due to camera optical axis and road surface out of plumb bring shooting angle error (referring to Fig. 8)；
900. calculate the construction depth of bituminous pavement.
In the above-mentioned steps 100 of this method, choose left and right two video cameras specification it is identical, imaging surface parallel co-planar and
Row is aligned, and the video camera between left and right at a distance constructs the world as origin using measuring table and sit as binocular camera
Mark system, binocular camera is demarcated using Zhang Zhengyou calibration method, solve two video cameras internal reference and outer ginseng.It needs to illustrate
, the internal reference of video camera is only related with the specification of video camera, just uniquely determines in video camera factory, joins only later outside video camera
Relative positional relationship between two video cameras in left and right is related.Video camera internal reference contains focal length of camera f, scale factor
zc, video camera principle point location u0、v0, reacted the internal structure situation of video camera.Join outside video camera contain video camera relative to
Relative to tessellated translation vector T, the relative position reacted between video camera is closed for tessellated spin matrix R and video camera
System.
Further, video camera is demarcated, using Zhang Zhengyou calibration method, shoots one group of chessboard trrellis diagram using video camera
Picture, chessboard table images are established as shown in Fig. 2, then identify the X-comers in digital picture using image recognition technology
X-comers in digital picture and the corresponding relationship between X-comers in the real world, solve the interior of video camera
Ginseng and outer ginseng.Include the following steps:
101. setting the coordinate of the next point of world coordinate system as P (X, Y, Z), the pixel coordinate of any on corresponding image
For p (u, v), then the conversion process of world coordinates to pixel coordinate carries out as the following formula:
ax=zcJ；
ay=zcf；
Wherein, K is the internal reference matrix of video camera, u0、v0For the pixel coordinate of video camera principal point, ax、ayFor focal length of camera
Parameter, R are video camera relative to tessellated 3 × 3 spin matrix, and T is video camera relative to tessellated 3 × 1 translation vector,
zcFor scale factor, f is focal length of camera.
102. assuming that camera coordinate system and world coordinate system overlap, then world coordinate system is located at the plane of Z=0
On, Z=0, above formula conversion can be enabled are as follows:
H=K [r1 r2T]；
In formula, the homography matrix that H is 3 × 3, r1, r2The respectively first row and secondary series of camera rotation matrix R.
Whole video camera internal references and outer ginseng are contained in homography matrix, and homography matrix H is write and does three column vectors
Form [h1 h2 h3], using coordinate convert in constraint condition, solve homography matrix according to the following formula；
103. shooting one group of chessboard table images using binocular camera, gridiron pattern then is identified using image recognition technology
Angle point, by under pixel coordinate system angle point p (u, v) and world coordinate system under angle point P (X, Y, Z) be used as given value, calculate
Homography matrix solves whole video camera internal references and outer ginseng.
In the above-mentioned steps 200 of this method, as shown in figure 3, since the specification that will control two video cameras 1 is identical, at
Image planes parallel co-planar and row alignment, video camera 1 at a distance is used as one group of binocular camera between left and right.Namely
It says, the video camera 1 of left and right two in the present invention constitutes one group of binocular camera, by binocular camera with certain altitude right angle setting
In on bituminous pavement 2, two video cameras 1 in left and right are controlled by computer and are taken pictures simultaneously, utilizes two width of left and right number obtained
Image detects 2 construction depth of bituminous pavement, and the part that two 1 shooting areas of video camera in left and right overlap is tested
Region 3.The video camera that binocular camera must select focal length fixed cannot select the video camera mirror with autozoom function
Head.It should be noted that two video cameras 1 in left and right are mounted on 2 top of bituminous pavement, the optical axis of video camera 1 by certain altitude
Orthogonal with bituminous pavement 2, the shooting area of two video cameras 1 in left and right overlaps and (refers to Fig. 3).
Further include following steps in the above-mentioned steps 300 of this method:
301. are pressed left color image and right color image by red (R), green (G), blue three channels (B) respectively
After formula calculates, the left single channel gray level image and right single channel gray level image that are converted to；
F (x, y)=R (x, y) × 0.299+G (x, y) × 0.587+B (x, y) × 0.114；
Wherein, f (x, y) is the gray value of pixel, and R (x, y), G (x, y), B (x, y) are respectively the red, green of pixel
Color, blue three channels value.
302. carry out denoising to left single channel gray level image and right single channel gray level image using median filtering, obtain
Left gray level image and right gray level image (referring to Fig. 4).
Median filtering denoising refers to be slided on the image using 3 × 3 two-dimentional sleiding form of square, will be to
The gray value of processing is placed on the centre of window, and gray value all in window is arranged from small to large, is calculated in window
The intermediate value of gray value then determines that this gray value is different when gray value to be processed is equal to the maximum value or minimum value of gray value
Often, gray value to be processed is replaced with the intermediate value of gray value, exports replaced gray value；It is on the contrary then be determined as normal value, it exports
The gray value of script.
Further include following steps in the above-mentioned steps 400 of this method:
401. determine distortion factor k according to the internal reference of video camera according to the following formula1、k2:
Wherein, k1、k2For the distortion factor of video camera, u, v are distortionless pixel coordinate, and x, y are distortionless continuous image
Plain coordinate, u0、v0For the pixel coordinate of video camera principal point,For the pixel coordinate after distortion.It should be noted that distortion
Correction, which refers to, is corrected barrel distortion issuable in image or pincushion distortion, and the foundation of correction is the interior of video camera
Ginseng.
402. utilize resulting distortion of camera coefficient k1、k2As the following formula, respectively to left gray level image and right gray level image into
Line distortion correction:
In the above-mentioned steps 500 of this method, three-dimensional correction refers to that the relative position to two video cameras in left and right carries out school
Just, two video cameras in left and right can have error during the installation process, and two imaging planes can not be substantially parallel coplanar and row pair
It is quasi-, it is therefore desirable to which that two images are subjected to three-dimensional correction.Join with Bouguet algorithm using outside video camera obtained by calibrating to image
It is corrected.Further include following steps:
501. determine the relative positional relationship between two video cameras in left and right, and formula is as follows:
R=RrRlT；
T=Tr-RTl；
Wherein, R is 3 × 3 relative rotation matrices between two video cameras in left and right, and T is 3 between two video cameras in left and right
× 1 relative translation vector, Rl、RrRespectively two video cameras in left and right are relative to tessellated 3 × 3 spin matrix, Tl、TrRespectively
It is two video cameras in left and right relative to tessellated 3 × 1 translation vector；
502. are converted using Douglas Rodríguez, and relative rotation matrices are decomposed into left image and the respective synthesis of right image
Spin matrix rl、rr；
503. calculate the respective spin matrix R of left and right two imageslt、Rrt, by left image according to spin matrix RlRotation, will
Right image is according to spin matrix RrIt is rotated, keeps the polar curve of two images horizontal, and pole is at infinity, complete three-dimensional school
Just.Formula is as follows:
Rlt=Rr ectrl；
Rrt=Rr ectrr；
Rr ect=[e1 e2 e3]；
e3=e1×e2；
In formula, Rlt、RrtFor respective 3 × 3 spin matrix of left images respectively, R is 3 between two video cameras in left and right
× 3 relative rotation matrices, T are 3 × 1 relative translation vectors between two video cameras in left and right, rl、”rrRespectively left images
Synthesize spin matrix.It should be noted that the image after correction refers to Fig. 5.
Further include following steps in the above-mentioned steps 600 of this method:
After 601. using each pixel on half global registration algorithm (SGBM) traversal image, the second correction left side is identified
The same pixel on gray level image and the second right gray level image of correction, and calculate the parallax value d of pixel, unit mm；
Wherein, xl、xrThe respectively same pixel is on the left gray level image of the second correction and the second right gray level image of correction
Pixel horizontal axis coordinate, zcFor scale factor；
602. calculate height value z of each pixel apart from camera plane with following formula；
Wherein, TxFor component of the relative translation vector T in X direction, unit mm indicates two video cameras in left and right
The distance between, f is focal length of camera, and unit mm, d are parallax value, unit mm；As it can be seen that parallax value d is bigger, then pixel
Point is closer apart from video camera, and parallax value d is smaller, then pixel is remoter apart from video camera.
The height value of all pixels point is formed a model matrix M by 603., recovers the threedimensional model on road surface, wherein mould
Type matrix M is one and contains the matrix of the pixel coordinate of each pixel and its corresponding pixel height value on image.
Fig. 6 is referred to, Fig. 6 is the tested region road surface model figure comprising Stereo matching error value of generation after the completion of Stereo matching.
Further include following steps in the above-mentioned steps 700 of this method:
701. are calculated as follows the single order difference coefficient of the corresponding height value of pixel, determine the position of Stereo matching erroneous pixel point
It sets；
K=(zi+1-zi)/(xi+1-xi)；
Wherein, k is the single order difference coefficient of pixel, xi, xi+1For i-th and the pixel abscissa of i+1 pixel, zi, zi+1
For i-th and the height value of i+1 pixel, unit mm.
702. point by single order difference coefficient greater than 1 is defined as the point of matching error, using 7 × 7 filter window, matching is wrong
Point accidentally is placed on window center, and pixel height value all in window is arranged from small to large, calculates picture in window
The intermediate value of vegetarian refreshments height value exports replaced pixel height value, be eliminated solid with the value of intermediate value replacement matching error
Road surface model figure after the value of matching error, adjust model display scale after it is shown in Figure 7.
Further include following steps in the above-mentioned steps 800 of this method:
The pixel height value of 801. couples of model matrix M does plane fitting, and the parameter a of fit Plane is calculated as follows out1,
a2, a3；
Wherein, xi, yiFor the pixel coordinate of ith pixel point, ziFor the height value of ith pixel point, unit mm, n are
Total pixel number in matrix.
802. calculate the height value of revised each pixel using following formula, complete shooting angle error correction；
hi=zi-a1xi-a2yi-a3；
Wherein, zi、hiHeight value respectively before shooting angle amendment with revised ith pixel point, unit mm,
Revised road surface model figure is obtained, it is shown in Figure 8.
In the above-mentioned steps 900 of this method, the construction depth H for calculating bituminous pavement is carried out as the following formulap, unit mm；
Wherein, hmaxFor the maximum value of pixel height value, unit mm, hiFor the height value of ith pixel point, unit
The line number and columns for being model matrix M for mm, m and n.
To verify effectiveness of the invention, the detection of bituminous pavement construction depth is carried out using the present invention and to 30 of acquisition
Measuring point image information carries out analytical calculation, calculated result is compared with the testing result for sanding method by hand, as a result such as 1 institute of table
Show.
Table 1:
From table 1 it follows that the test result maximum relative error of 30 measuring points is -8.45%, average relative error
It is 3.04%, related coefficient 0.933, detection method of the invention is smaller compared to the testing result error for sanding method by hand, phase
Pass degree is higher, has good detection effect.
The present invention is due to using two video cameras in left and right, by acquiring bituminous pavement respectively to two video cameras in left and right
Left color image and right color image successively carry out gray proces, distortion correction, three-dimensional correction, the three-dimensional mould for recovering road surface
Type, the value for eliminating Stereo matching mistake, the shooting angle error for correcting video camera, finally calculate the construction depth of bituminous pavement；
It is illuminated by the light in the detection process smaller with the influence of road surface self color, is not only capable of measuring bituminous pavement construction depth, simultaneously
The threedimensional model on road surface can also be recovered, can more intuitively reflect the technology status information of bituminous pavement, for testing staff
With reference to；It overcomes laser texture meter method and needs disadvantage using special equipment and expensive, taken the photograph using a pair of common
Detection can be completed in camera lens；It overcomes traditional-handwork and sands method and the electronic method detection speed that sands relatively slowly and by artificial subjectivity
The shortcomings that being affected, have many advantages, such as rapidly and efficiently, be not easy to be disturbed, price economy and testing result it is more accurate.
It should be noted that explaining in detail for the respective embodiments described above, purpose, which is only that, solves the present invention
It releases, in order to be able to preferably explain the present invention, still, these descriptions cannot be with any explanation at being to limit of the invention
System, in particular, each feature described in various embodiments can also mutual any combination, to form other implementations
Mode, in addition to there is clearly opposite description, these features should be understood to can be applied in any one embodiment, and simultaneously
It is not only limited to described embodiment.
Claims (10)
1. a kind of bituminous pavement construction depth detection method based on binocular vision, which comprises the steps of:
100. internal reference and the outer ginseng of acquisition left and right two video camera；
200. acquire the left color image and right color image of bituminous pavement using two video cameras in left and right respectively；
Left color image and right color image are processed into left gray level image and right gray level image by 300. respectively；
400. obtain according to the internal reference of two video cameras in left and right respectively to left gray level image and right gray level image progress distortion correction
To the left gray level image of the first correction and the first right gray level image of correction；
500. internal references according to the video camera of left and right two and outer ginseng, respectively to the left gray level image of the first correction and the first right ash of correction
It spends image and carries out three-dimensional correction, obtain the left gray level image of the second correction and the second right gray level image of correction；
The left gray level image of 600. pair of second correction and the second right gray level image of correction carry out Stereo matching, identify the second correction left side
Corresponding pixel on gray level image and the second right gray level image of correction, calculates parallax value d, is calculated on image according to parallax value d
Each pixel, apart from the height value of camera plane, generates one and includes each pixel on image under camera coordinate system
Pixel coordinate and its corresponding height value information model matrix M, recover the threedimensional model on road surface；
Threshold value is arranged in the difference coefficient value of height value between two neighboring pixel in 700. pairs of model matrixs, determines Stereo matching mistake
The accidentally position of value, is modified using value of the median filtering window to three-dimensional matching error, eliminates the value of Stereo matching mistake；
800. couples of model matrix M carry out plane fittings, and model matrix M and fit Plane are subtracted each other, when amendment acquisition image due to
Camera optical axis bring shooting angle error not exclusively vertical with road surface；
900. calculate the construction depth of bituminous pavement.
2. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 300:
301. are counted left color image and right color image as the following formula by red (R), green (G), blue three channels (B) respectively
After calculation, left single channel gray level image and right single channel gray level image are converted to；
F (x, y)=R (x, y) × 0.299+G (x, y) × 0.587+B (x, y) × 0.114；
Wherein, f (x, y) is the gray value of pixel, R (x, y), G (x, y), B (x, y) be respectively the red of pixel, green,
The value in blue three channels.
3. the bituminous pavement construction depth detection method according to claim 2 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 300:
302. carry out denoising to left single channel gray level image and right single channel gray level image using median filtering, obtain left ash
Spend image and right gray level image.
4. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 400:
401. determine distortion factor k according to the internal reference of video camera according to the following formula1、k2:
Wherein, k1、k2For the distortion factor of video camera, u, v are distortionless pixel coordinate, and x, y are distortionless contiguous pixels
Coordinate, u0、v0For the pixel coordinate of video camera principal point,For the pixel coordinate after distortion；
402. utilize resulting distortion of camera coefficient k1、k2As the following formula, left gray level image and right gray level image are carried out respectively abnormal
Become correction:
5. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 500:
501. determine the relative positional relationship between two video cameras in left and right:
502. are converted using Douglas Rodríguez, and relative rotation matrices are decomposed into left image and the respective synthesis of right image rotates
Matrix rl、rr；
503. calculate the respective spin matrix R of left and right two imageslt、Rrt, by left image according to spin matrix RlRotation, by right figure
As according to spin matrix RrIt is rotated, keeps the polar curve of two images horizontal, and pole is at infinity, complete three-dimensional correction.
6. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 600:
After 601. using each pixel on half global registration algorithm (SGBM) traversal image, the left gray scale of the second correction is identified
The same pixel on image and the second right gray level image of correction, and calculate the parallax value d of pixel；
Wherein, xl、xrPicture of the respectively same pixel on the left gray level image of the second correction and the second right gray level image of correction
Plain horizontal axis coordinate, zcFor scale factor；
602. calculate height value z of each pixel apart from camera plane with following formula；
Wherein, TxFor component of the relative translation vector T in X direction, unit mm is indicated between two video cameras in left and right
Horizontal distance, f are focal length of camera, and unit mm, d are parallax value, unit mm；
The height value of all pixels point is formed a model matrix M by 603., recovers the threedimensional model on road surface, wherein model square
Battle array M is one and contains the matrix of the pixel coordinate of each pixel and its corresponding pixel height value on image.
7. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 700:
701. are calculated as follows the single order difference coefficient of the corresponding height value of neighbor pixel, determine the position of Stereo matching erroneous pixel point
It sets；
K=(zi+1-zi)/(xi+1-xi)；
Wherein, k is the single order difference coefficient of pixel, xi, xi+1For i-th and the pixel abscissa of i+1 pixel, zi, zi+1It is
The height value of i and i+1 pixel, unit mm；
702. point by single order difference coefficient greater than 1 is defined as the point of matching error, using 7 × 7 filter window, matching error
Point is placed on window center, and the height value of all the points in window is arranged from small to large, is calculated in window inner height value
Value exports replaced pixel height value with the value of intermediate value replacement matching error.
8. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 800:
The pixel height value of 801. couples of model matrix M does plane fitting, and the parameter a of fit Plane is calculated as follows out1, a2,
a3；
Wherein, xi, yiFor the pixel coordinate of ith pixel point, ziFor the height value of ith pixel point, n is picture total in matrix
Vegetarian refreshments number.
9. the bituminous pavement construction depth detection method according to claim 8 based on binocular vision, which is characterized in that
Further include following steps in above-mentioned steps 800:
802. calculate the height value of revised each pixel using following formula, complete shooting angle error correction；
hi=zi-a1xi-a2yi-a3；
Wherein, zi、hiHeight value respectively before shooting angle amendment with revised ith pixel point, unit mm.
10. the bituminous pavement construction depth detection method according to claim 1 based on binocular vision, which is characterized in that
In above-mentioned steps 900, the construction depth H for calculating bituminous pavement is carried out as the following formulap, unit mm；
Wherein, hmaxFor the maximum value of pixel height value, unit mm, hiFor the height value of ith pixel point, unit mm, m
The line number and columns for being model matrix M with n.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910053244.9A CN109919856B (en) | 2019-01-21 | 2019-01-21 | Asphalt pavement structure depth detection method based on binocular vision |
| CN202310309976.6A CN116342674A (en) | 2019-01-21 | 2019-01-21 | Method for calculating asphalt pavement construction depth by three-dimensional model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910053244.9A CN109919856B (en) | 2019-01-21 | 2019-01-21 | Asphalt pavement structure depth detection method based on binocular vision |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310309976.6A Division CN116342674A (en) | 2019-01-21 | 2019-01-21 | Method for calculating asphalt pavement construction depth by three-dimensional model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109919856A true CN109919856A (en) | 2019-06-21 |
| CN109919856B CN109919856B (en) | 2023-02-28 |
Family
ID=66960505
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910053244.9A Active CN109919856B (en) | 2019-01-21 | 2019-01-21 | Asphalt pavement structure depth detection method based on binocular vision |
| CN202310309976.6A Pending CN116342674A (en) | 2019-01-21 | 2019-01-21 | Method for calculating asphalt pavement construction depth by three-dimensional model |
Family Applications After (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310309976.6A Pending CN116342674A (en) | 2019-01-21 | 2019-01-21 | Method for calculating asphalt pavement construction depth by three-dimensional model |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN109919856B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111091063A (en) * | 2019-11-20 | 2020-05-01 | 北京迈格威科技有限公司 | Living body detection method, device and system |
| CN111553878A (en) * | 2020-03-23 | 2020-08-18 | 四川公路工程咨询监理有限公司 | Method for detecting paving uniformity of asphalt pavement mixture based on binocular vision |
| CN111862234A (en) * | 2020-07-22 | 2020-10-30 | 中国科学院上海微系统与信息技术研究所 | Binocular camera self-calibration method and system |
| CN112819820A (en) * | 2021-02-26 | 2021-05-18 | 大连海事大学 | Pavement asphalt repair detection method based on machine vision |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117649454A (en) * | 2024-01-29 | 2024-03-05 | 北京友友天宇系统技术有限公司 | Binocular camera external parameter automatic correction method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090305820A1 (en) * | 2007-09-18 | 2009-12-10 | Scott Denton | Golf gps device |
| CN102635056A (en) * | 2012-04-01 | 2012-08-15 | 长安大学 | Measuring method for construction depth of asphalt road surface |
| CN104775349A (en) * | 2015-02-15 | 2015-07-15 | 云南省交通规划设计研究院 | Tester and measuring method for structural depth of large-porosity drainage asphalt pavement |
| CN105205822A (en) * | 2015-09-21 | 2015-12-30 | 重庆交通大学 | Real-time detecting method for asphalt compact pavement segregation degree |
| CN105225482A (en) * | 2015-09-02 | 2016-01-06 | 上海大学 | Based on vehicle detecting system and the method for binocular stereo vision |
| CN106845424A (en) * | 2017-01-24 | 2017-06-13 | 南京大学 | Road surface remnant object detection method based on depth convolutional network |
-
2019
- 2019-01-21 CN CN201910053244.9A patent/CN109919856B/en active Active
- 2019-01-21 CN CN202310309976.6A patent/CN116342674A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090305820A1 (en) * | 2007-09-18 | 2009-12-10 | Scott Denton | Golf gps device |
| CN102635056A (en) * | 2012-04-01 | 2012-08-15 | 长安大学 | Measuring method for construction depth of asphalt road surface |
| CN104775349A (en) * | 2015-02-15 | 2015-07-15 | 云南省交通规划设计研究院 | Tester and measuring method for structural depth of large-porosity drainage asphalt pavement |
| CN105225482A (en) * | 2015-09-02 | 2016-01-06 | 上海大学 | Based on vehicle detecting system and the method for binocular stereo vision |
| CN105205822A (en) * | 2015-09-21 | 2015-12-30 | 重庆交通大学 | Real-time detecting method for asphalt compact pavement segregation degree |
| CN106845424A (en) * | 2017-01-24 | 2017-06-13 | 南京大学 | Road surface remnant object detection method based on depth convolutional network |
Non-Patent Citations (2)
| Title |
|---|
| AKHTAR HANIF等: ""Inflight helicopter blade track measurement using computer vision"", 《2014 IEEE REGION 10 SYMPOSIUM》 * |
| 何力: ""基于数字图像技术的沥青混凝土构造深度检测研究"", 《北方交通》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111091063A (en) * | 2019-11-20 | 2020-05-01 | 北京迈格威科技有限公司 | Living body detection method, device and system |
| CN111091063B (en) * | 2019-11-20 | 2023-12-29 | 北京迈格威科技有限公司 | Living body detection method, device and system |
| CN111553878A (en) * | 2020-03-23 | 2020-08-18 | 四川公路工程咨询监理有限公司 | Method for detecting paving uniformity of asphalt pavement mixture based on binocular vision |
| CN111862234A (en) * | 2020-07-22 | 2020-10-30 | 中国科学院上海微系统与信息技术研究所 | Binocular camera self-calibration method and system |
| CN111862234B (en) * | 2020-07-22 | 2023-10-20 | 中国科学院上海微系统与信息技术研究所 | Binocular camera self-calibration method and system |
| CN112819820A (en) * | 2021-02-26 | 2021-05-18 | 大连海事大学 | Pavement asphalt repair detection method based on machine vision |
| CN112819820B (en) * | 2021-02-26 | 2023-06-16 | 大连海事大学 | Road asphalt repairing and detecting method based on machine vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109919856B (en) | 2023-02-28 |
| CN116342674A (en) | 2023-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109919856A (en) | Bituminous pavement construction depth detection method based on binocular vision | |
| CN112669393B (en) | Laser radar and camera combined calibration method | |
| CN110285793B (en) | Intelligent vehicle track measuring method based on binocular stereo vision system | |
| CN102364299B (en) | Calibration technology for multiple structured light projected three-dimensional profile measuring heads | |
| CN104240262B (en) | Calibration device and calibration method for outer parameters of camera for photogrammetry | |
| CN101692283B (en) | Method for on-line self-calibration of external parameters of cameras of bionic landing system of unmanned gyroplane | |
| CN106978774B (en) | A kind of road surface pit slot automatic testing method | |
| CN107025670A (en) | A kind of telecentricity camera calibration method | |
| CN109443245B (en) | Multi-line structured light vision measurement method based on homography matrix | |
| CN110057295A (en) | It is a kind of to exempt from the monocular vision plan range measurement method as control | |
| CN110375648A (en) | The spatial point three-dimensional coordinate measurement method that the single camera of gridiron pattern target auxiliary is realized | |
| CN101216296A (en) | Binocular vision rotating axis calibration method | |
| CN103994732B (en) | A kind of method for three-dimensional measurement based on fringe projection | |
| CN106023193B (en) | A kind of array camera observation procedure detected for body structure surface in turbid media | |
| CN106447733B (en) | Method, system and device for determining cervical vertebra mobility and moving axis position | |
| CN111091076B (en) | Tunnel limit data measuring method based on stereoscopic vision | |
| CN104568963A (en) | Online three-dimensional detection device based on RGB structured light | |
| CN109859269B (en) | Shore-based video auxiliary positioning unmanned aerial vehicle large-range flow field measuring method and device | |
| CN110966956A (en) | Binocular vision-based three-dimensional detection device and method | |
| CN112305576A (en) | Multi-sensor fusion SLAM algorithm and system thereof | |
| CN107610183A (en) | New striped projected phase height conversion mapping model and its scaling method | |
| CN110047111A (en) | A kind of airplane parking area shelter bridge butting error measurement method based on stereoscopic vision | |
| CN111009030A (en) | Multi-view high-resolution texture image and binocular three-dimensional point cloud mapping method | |
| CN112348775A (en) | Vehicle-mounted all-round-looking-based pavement pool detection system and method | |
| CN110044266B (en) | Photogrammetry system based on speckle projection |
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 |