CN110097494A - A kind of cargo localization method based on Fourier-Mellin transform - Google Patents
A kind of cargo localization method based on Fourier-Mellin transform Download PDFInfo
- Publication number
- CN110097494A CN110097494A CN201910345255.4A CN201910345255A CN110097494A CN 110097494 A CN110097494 A CN 110097494A CN 201910345255 A CN201910345255 A CN 201910345255A CN 110097494 A CN110097494 A CN 110097494A
- Authority
- CN
- China
- Prior art keywords
- image
- cargo
- shelf
- fourier
- current
- 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
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000004807 localization Effects 0.000 title claims abstract description 24
- 230000009466 transformation Effects 0.000 claims abstract description 24
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims abstract description 20
- 230000007246 mechanism Effects 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 8
- 238000001228 spectrum Methods 0.000 claims description 23
- 230000033001 locomotion Effects 0.000 claims description 13
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000003384 imaging method Methods 0.000 claims description 2
- 238000009434 installation Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 4
- 238000001914 filtration Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0833—Tracking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/14—Transformations for image registration, e.g. adjusting or mapping for alignment of images
- G06T3/147—Transformations for image registration, e.g. adjusting or mapping for alignment of images using affine transformations
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of cargo localization method based on Fourier-Mellin transform, comprising: step 1 establishes absolute coordinate system, and calibration point, the location information of identification calibration point location shelf is arranged;Step 2, installation camera are being protected from light place;Step 3 establishes characteristics of image library, characteristic image of the particular frame image of the cargo of extract real-time shooting as corresponding cargo location, pass through the location information of shelf, affine transformation matrix is obtained, then affine transformation obtains new characteristic image, is denoted as cargo characteristic image library;Step 4, obtains the cargo image for currently taking object locations of structures to shoot, and the image registration based on Fourier-Mellin transform calculates the position orientation relation of cargo and Qu Wu mechanism;Step 5 is updated pattern image library, and jumps to step 4 and carry out subsequent image registration work;This method realizes the high accuracy positioning of a variety of shelf and cargo, the image information for the calibration that timely updates, applied widely and facilitate operation.
Description
Technical field
The present invention relates to mobile robot technology field, especially a kind of mobile machine based on Fourier-Mellin transform
People's cargo localization method.
Background technique
In logistic storage field, there are more and more robots to be used for the carrying and sorting of cargo.Utilize machine vision
Carrying out positioning to object and robot is the hot spot in current mobile robot field.With machine man-based development, applied field
Scape is further extensive.
In this industrial circle, an important process is grabbed to the specific cargo on shelf, only
There is accurate positioning to cargo, is just able to achieve the correct crawl to cargo.Mainly have to the positioning method of cargo at present following several
Kind: (1) cargo according to certain rules put on shelf, by the relative position of control robot and shelf, realizes that cargo is fixed
Position, the positioning and control precision that this method excessively relies on robot are likely to result in picking mistake when deviation occurs in control
Accidentally;(2) by the cargo pallet of customization, the pallet bigger than cargo is often used, picking is improved by bigger area
Serious forgiveness, it is apparent that this method reduces the utilization rate of shelf, cost is also higher;(3) additional label is added on cargo,
Such as two dimensional code, but the cargo having cannot add label or addition label after discrimination it is lower, limit the application of this method
Scene;Market need it is a kind of do not need to carry out cargo additional processing, cargo positioning side that is applied widely and facilitating operation
Method, the present invention are used to solve such problems.
Summary of the invention
To solve the deficiencies in the prior art, the purpose of the present invention is to provide a kind of shiftings based on Fourier-Mellin transform
Mobile robot cargo localization method marks shelf with the mark point on ground or shelf, then passes through Fourier-Mellin transform pair
Cargo image is calculated the high accuracy positioning, it can be achieved that a variety of shelf and cargo, while every time by will be updated when calibration point
The image information of calibration, reach adaptively surface wear, variation situations such as, do not need to carry out cargo additional processing, be applicable in
Range is wide and facilitates operation.
In order to achieve the above objectives, the present invention adopts the following technical scheme that:
A kind of cargo localization method based on Fourier-Mellin transform, includes the following steps:
Step 1 establishes absolute coordinate system,
Before moveable robot movement,
Multiple calibration points are artificially set in coordinate system, and calibration point is arranged in movement routine for marking the position of shelf
Set, the absolute coordinate of each calibration point be it is known that positional relationship between mark point and shelf it is also known that;
It identifies calibration point, positions the location information of shelf;
Step 2 is protected from light place for what camera was mounted on that mobile robot takes object structure;
Step 3 establishes characteristics of image library,
Cargo on shelf is shot, spy of the particular frame image as corresponding cargo location in extract real-time video
Image is levied, and then forms a series of characteristic image of cargo on shelf;
The position orientation relation that current robot and shelf are obtained by label point image, calculates affine matrix, will obtain before
Characteristic image new characteristic image is obtained by affine transformation, be stored in cargo characteristic image library;
Step 4 obtains the cargo image for currently taking object locations of structures to shoot when taking object structure to start to position cargo,
If current signature image texture and the non-overlapping place in characteristic image library, carried out with the image of previous frame known location
Image registration based on Fourier-Mellin transform, currently to be taken the relative pose of object structure and cargo;
If current signature image texture has overlapping with characteristic image library, the two become based on Fourier-plum forests
The image registration changed, to obtain the relative pose of current cargo;
Current location characteristic image is registrated, after the completion of registration, by affine transformation by step 5 with characteristic image library
Current position image afterwards is mixed with characteristic image, and the new images for obtaining having both two characteristics of image are as new known bits
Image is set, the pattern image nearest apart from present frame is deleted, carries out the update in pattern image library, and jumps to step 4 progress
Subsequent cargo positions work.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned identifies calibration point in step 1, fixed
The specific method of location information of position shelf includes: that calibration point is fixed on ground or shelf, calibration point and shelf it is opposite
Position be it is fixed and known, mobile robot is imaged calibration point by camera, identifies that calibration point determines shelf
Number.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned, which is characterized in that in step 1, mark
The representation of fixed point includes: one-dimension code, two dimensional code, additional character or special texture.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned,
In step 2, place is protected from light by what camera was mounted on that mobile robot takes object structure;If the camera can not cover
Cover all models
It encloses, then shelf is divided into multiple subregions, each subregion has movement of the mark point for robot.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned,
In step 3, characteristics of image library is established,
Cargo is placed on shelf, the cargo on shelf is shot, the particular frame image in extract real-time video is made
For the characteristic image of corresponding cargo location, and then form a series of characteristic image of cargo on shelf;Particular frame be extracted according to
According to being that first frame is key frame, the picture frame when acquisition time with previous frame is more than regular hour T is key frame, and upper
The picture frame when acquisition position of one frame is more than certain range M is key frame;
By the identification to mark point, the location information of shelf is obtained, affine transformation matrix is calculated, utilizes affine transformation square
Battle array carries out affine transformation to the image of the shot by camera of Qu Wu mechanism and obtains new characteristic image, is stored in cargo characteristic image
Library;The location information of shelf mainly includes the relative position and relative angle between mobile robot and shelf;
The image of shooting need to cover the whole region of current shelf, if can not to cover shelf complete for the coverage area of Qu Wu mechanism
Multiple calibration points are placed in portion region before shelf, and shelf are divided into multiple subregions, cooperate the mobile realization of robot to shelf
The covering in region.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned,
In step 4, when taking object structure to start to position cargo, the cargo figure for currently taking object locations of structures to shoot is obtained
Picture,
If current signature image texture and the non-overlapping place in characteristic image library, carried out with the image of previous frame known location
Image registration based on Fourier-Mellin transform, the relative pose with cargo currently to be taken object structure;
If current signature image texture have with characteristic image library it is overlapping, to current signature image and the characteristic pattern for having overlapping
As library image carries out the image registration based on Fourier-Mellin transform, to obtain the relative pose of current cargo;
The method for obtaining transverse shifting relationship between cargo and load-engaging device includes: the image arrived using current shooting and
The cross energy spectrum for knowing the image Fourier variation of position, solves the translation relation of image, and then releases cargo and take object structure
Positional relationship;Assuming that known image and the Fourier spectrum of present image are F1And F2, then the cross energy of two images is composed in frequency domain
Are as follows:
Cross energy is composed and carries out inverse transformation, peak value is obtained and is located at (x0,y0) impulse function, (x0,y0) it is present image
It is calculated cargo by taking object structure initial position and camera internal reference with the position deviation of known image and the lateral of Qu Wu mechanism is moved
Dynamic relationship.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned,
In step 4, when taking object structure to start to position cargo, the cargo figure for currently taking object locations of structures to shoot is obtained
Picture,
If current signature image texture and the non-overlapping place in characteristic image library, base is carried out with the image of previous frame known location
In the image registration of Fourier-Mellin transform, currently to be taken the rotation relationship of object structure;
If current signature image texture has overlapping with characteristic image library, the two become based on Fourier-plum forests
The image registration changed, to obtain the rotation relationship of current cargo;
The method for obtaining rotation angle between cargo and load-engaging device includes: the image for arriving current shooting and known location
The energy conversion of image Fourier spectrum solve the rotation relationship of image to polar coordinate system, and then release cargo and take object structure
Rotation relationship;
Assuming that known image and the Fourier spectrum of present image are F1And F2, the corresponding energy spectrum of Fourier is calculated, formula is such as
Under:
Energy is converted to polar coordinate system M1(ρ, θ) and M2(ρ, θ), to M under polar coordinates1And M2It is related to solve phase
It obtains being located at θ0The angle of impulse function at position, i.e. present image and known image is θ0;
By taking object structure initial position and camera internal reference, the positional relationship of cargo and Qu Wu mechanism is calculated.
A kind of cargo localization method based on Fourier-Mellin transform above-mentioned,
In step 4, when taking object structure to start to position cargo, the cargo figure for currently taking object locations of structures to shoot is obtained
Picture,
If current signature image texture and the non-overlapping place in characteristic image library, base is carried out with the image of previous frame known location
In the image registration of Fourier-Mellin transform, currently to be taken the distance relation of object structure;
If current signature image texture has overlapping with characteristic image library, the two become based on Fourier-plum forests
The image registration changed, to obtain the distance relation of current cargo;
The method for obtaining distance between cargo and load-engaging device includes: the image for arriving current shooting and the figure of known location
As the energy conversion of Fourier spectrum solves the proportionate relationship of image to logarithmic coordinates system, and then it is implemented in combination with the Hash table of foundation
Distance representated by current ratio is inquired, and then obtains cargo and takes at a distance from object structure;Assuming that known image and current figure
The Fourier spectrum of picture is F1(ξ, η) and F2They are converted into logarithmic coordinates system, F by (ξ, η)1(log ξ, log η) and F2(logξ,
Log η), phase correlation is solved, formula is as follows:
The scale factor s of present image and known image is obtained, the Hash for pre-establishing scale factor and actual distance is looked into
Ask table, can be obtained under current state cargo with take object it is inter-agency at a distance from.
The invention has the beneficial effects that:
Shelf cargo localization method proposed by the present invention by calibration point make mobile robot obtain current shelf location and
Number, after determining robot before correct shelf, calculates cargo image by Fourier-Mellin transform, passes through
With the comparison in characteristic image library, lateral position, deflection angle and the distance of cargo and load-engaging device are obtained;Such method is to goods
The detection positioning accurate parasexuality of object is preferable, and calculation amount is small, effectively improves matching precision and performance, to improve to movement
Precision robot cargo positioning and picked and placed;
The present invention will be updated the image information of calibration when passing through calibration point every time, adaptively the feelings such as surface wear, variation
Condition does not need to carry out cargo additional processing, applied widely and facilitate operation.
Detailed description of the invention
Fig. 1 is a kind of flow chart of embodiment of the invention.
Specific embodiment
Specific introduce is made to the present invention below in conjunction with the drawings and specific embodiments.
A kind of cargo localization method based on Fourier-Mellin transform, includes the following steps:
Step 1 establishes absolute coordinate system,
Before moveable robot movement,
Multiple calibration points are artificially set in coordinate system, and calibration point is arranged in movement routine for marking the position of shelf
Set, the absolute coordinate of each calibration point be it is known that positional relationship between mark point and shelf it is also known that;
It identifies calibration point, positions the location information of shelf;
Identify calibration point, the specific method for positioning the location information of shelf includes: that calibration point is fixed on ground or shelf
On, the relative position of calibration point and shelf be it is fixed and known, mobile robot is imaged calibration point by camera,
Identification calibration point determines the number of shelf.
The representation of calibration point includes: one-dimension code, two dimensional code, additional character or special texture.
Step 2 is protected from light place for what camera was mounted on that mobile robot takes object structure;
Place is protected from light by what camera was mounted on that mobile robot takes object structure;If the camera can not cover all ranges,
Shelf are then divided into multiple subregions, each subregion has movement of the mark point for robot.
Step 3 establishes characteristics of image library,
Cargo is placed on shelf, the cargo on shelf is shot, the particular frame image in extract real-time video is made
For the characteristic image of corresponding cargo location, and then form a series of characteristic image of cargo on shelf;Particular frame be extracted according to
According to being that first frame is key frame, the picture frame when acquisition time with previous frame is more than regular hour T is key frame, and upper
The picture frame when acquisition position of one frame is more than certain range M is key frame;
By the identification to mark point, the location information of shelf is obtained, affine transformation matrix is calculated, utilizes affine transformation square
Battle array carries out affine transformation to the image of the shot by camera of Qu Wu mechanism and obtains new characteristic image, is stored in cargo characteristic image
Library;The location information of shelf mainly includes the relative position and relative angle between mobile robot and shelf;
The image of shooting need to cover the whole region of current shelf, if can not to cover shelf complete for the coverage area of Qu Wu mechanism
Multiple calibration points are placed in portion region before shelf, and shelf are divided into multiple subregions, cooperate the mobile realization of robot to shelf
The covering in region.
Step 4 obtains the cargo image for currently taking object locations of structures to shoot when taking object structure to start to position cargo,
If current signature image texture and the non-overlapping place in characteristic image library, carried out with the image of previous frame known location
Image registration based on Fourier-Mellin transform, currently to be taken the relative pose of object structure and cargo;
If current signature image texture has overlapping with characteristic image library, the two become based on Fourier-plum forests
The image registration changed, to obtain the relative pose of current cargo;
There are three parts for cargo and the position orientation relation of Qu Wu mechanism, and one is transverse shifting, and one is angle, and one is ruler
Degree separately solves.
To obtain the positional relationship of cargo Yu Qu Wu mechanism, then step 4 are as follows:
In step 4, when taking object structure to start to position cargo, the cargo figure for currently taking object locations of structures to shoot is obtained
Picture,
If current signature image texture and the non-overlapping place in characteristic image library, carried out with the image of previous frame known location
Image registration based on Fourier-Mellin transform, the relative pose with cargo currently to be taken object structure;
If current signature image texture have with characteristic image library it is overlapping, to current signature image and the characteristic pattern for having overlapping
As library image carries out the image registration based on Fourier-Mellin transform, to obtain the relative pose of current cargo;
The method for obtaining transverse shifting relationship between cargo and load-engaging device includes: the image arrived using current shooting and
The cross energy spectrum for knowing the image Fourier variation of position, solves the translation relation of image, and then releases cargo and take object structure
Positional relationship;Assuming that known image and the Fourier spectrum of present image are F1And F2, then the cross energy of two images is composed in frequency domain
Are as follows:
Cross energy is composed and carries out inverse transformation, peak value is obtained and is located at (x0,y0) impulse function, (x0,y0) it is present image
It is calculated cargo by taking object structure initial position and camera internal reference with the position deviation of known image and the lateral of Qu Wu mechanism is moved
Dynamic relationship.
To obtain the rotation relationship of cargo Yu Qu Wu mechanism, then step 4 are as follows:
When taking object structure to start to position cargo, the cargo image for currently taking object locations of structures to shoot is obtained,
If current signature image texture and the non-overlapping place in characteristic image library, base is carried out with the image of previous frame known location
In the image registration of Fourier-Mellin transform, currently to be taken the rotation relationship of object structure;
If current signature image texture has overlapping with characteristic image library, the two become based on Fourier-plum forests
The image registration changed, to obtain the rotation relationship of current cargo;
The method for obtaining rotation angle between cargo and load-engaging device includes: the image for arriving current shooting and known location
The energy conversion of image Fourier spectrum solve the rotation relationship of image to polar coordinate system, and then release cargo and take object structure
Rotation relationship;
Assuming that known image and the Fourier spectrum of present image are F1And F2, the corresponding energy spectrum of Fourier are as follows:
Energy is converted to polar coordinate system M1(ρ, θ) and M2(ρ, θ), to M under polar coordinates1And M2It is related to solve phase
It obtains being located at θ0The angle of impulse function at position, i.e. present image and known image is θ0;
By taking object structure initial position and camera internal reference, the positional relationship of cargo and Qu Wu mechanism is calculated.
To obtain the distance relation of cargo Yu Qu Wu mechanism, then step 4 are as follows:
When taking object structure to start to position cargo, the cargo image for currently taking object locations of structures to shoot is obtained,
If current signature image texture and the non-overlapping place in characteristic image library, base is carried out with the image of previous frame known location
In the image registration of Fourier-Mellin transform, currently to be taken the distance relation of object structure;
If current signature image texture has overlapping with characteristic image library, the two become based on Fourier-plum forests
The image registration changed, to obtain the distance relation of current cargo;
The method for obtaining distance between cargo and load-engaging device includes: the image for arriving current shooting and the figure of known location
As the energy conversion of Fourier spectrum solves the proportionate relationship of image to logarithmic coordinates system, and then it is implemented in combination with the Hash table of foundation
Distance representated by current ratio is inquired, and then obtains cargo and takes at a distance from object structure;Assuming that known image and current figure
The Fourier spectrum of picture is F1(ξ, η) and F2They are converted into logarithmic coordinates system, F by (ξ, η)1(log ξ, log η) and F2(logξ,
Log η), phase correlation is solved, formula is as follows:
The scale factor s of present image and known image is obtained, the Hash for pre-establishing scale factor and actual distance is looked into
Ask table, can be obtained under current state cargo with take object it is inter-agency at a distance from.
Current location characteristic image is registrated, after the completion of registration, by affine transformation by step 5 with characteristic image library
Current position image afterwards is mixed with characteristic image, and the new images for obtaining having both two characteristics of image are as new known bits
Image is set, the pattern image nearest apart from present frame is deleted, carries out the update in pattern image library, and jumps to step 4 progress
Subsequent cargo positions work.
Identification method of the two dimensional code as calibration point is selected, object structure is taken to select mechanical arm, hardware platform selects zynq
7020 are specifically demonstrated:
(1) absolute coordinate system is established, before moveable robot movement, multiple calibration points are artificially set in coordinate system, is marked
Fixed point is arranged in movement routine, and for positioning the position of shelf, and the absolute coordinate of the calibration point of each artificial settings is equal
It is known.Such as, but not limited to: two-dimension code image is puted up on the ground on front side of shelf, the included information of two dimensional code is shelf volume
Number, angle is vertical with shelf, and robot is moved to right above two dimensional code, by the camera of robot bottom to two dimensional code into
Row imaging obtains shelf coding, obtains the angle theta between robot and shelfyaw。
(2) camera is mounted on mobile robot takes the front of object structure to be protected from light place, so that mobile robot is in shelf
It is preceding that the cargo on shelf can be shot automatically;Such as, but not limited to: control robot is moved to the goods of certain calibration point identification
Before frame;For mechanical arm in robot for grabbing cargo, the camera shot to cargo is installed at the place of being protected from light on mechanical arm;It is mechanical
Arm moving range can cover all cargo areas on shelf;If mechanical arm cannot cover all ranges, shelf are divided into several
Subregion, so that mechanical arm can cover all cargo ranges of each subregion.
(3) cargo on shelf is shot in advance, the particular frame image in extract real-time video is as corresponding cargo
The characteristic image of position, and then a series of characteristic image of cargo on shelf is formed, while carrying out affine transformation and obtaining new spy
Image is levied, cargo characteristic image library is denoted as;Such as, but not limited to: equally spaced figure is set in the cargo range of mechanical arm covering
As collection point, cargo is shot in collection point position, utilizes θyawAffine transformation matrix is constructed, by collected cargo figure
After carrying out affine transformation, store in the memory of robot with construction feature shape library.
(4) after the completion of the building of pattern image library, it is assumed that take object manipulator motion to somewhere, obtain Current mechanical arm camera
Captured cargo image, is denoted as I0;Four adjacent with current location are chosen in pattern image library according to Current mechanical arm position
Neighborhood characteristics figure I1、I2、I2And I4, object as registration;I0Respectively and I1、I2、I2And I4Become based on Fourier plum forests
The image registration changed;With I0And I1For, to I0And I1Fourier transformation is carried out, is F0(ε, η) and F1(ε,η)。
(5) to F0(ε, η) and F1(ε, η) carries out high-pass filtering, high-pass filtering template are as follows:
H (ε, η)=(1.0-X (ε, η)) (2.0-X (ε, η)),
Wherein
X (ε, η)=cos (π ε) cos (π η) -0.5≤ε, η≤0.5;
By F0(ε, η) and F1(ε, η) and template H (ε, η) is multiplied, and obtains filtered Fourier transform spectrum.
(6) by F0(ε, η) and F1(ε, η) is converted into polar coordinates F0(ρ, θ) and F1ρ radius is transformed to logarithm and sat by (ρ, θ)
ζ=log ρ is marked, F is obtained0(ζ, θ) and F1(ζ, θ), and then corresponding cross energy spectrum is obtained, formula is as follows:
Cross energy is composed and carries out Fourier inversion, formula is as follows:
The peak value of M ' is searched, if M ' has the peak value of multiple similar intensities or the intensity of peak value to be lower than threshold value, then it is assumed that I0
And I1Matching degree it is poor, continue to I2、I2And I4It is calculated, until finding qualified pattern image.
(7) if finding qualified characteristic image, such as I1, it is assumed that the peak position found is (ζ ', θ '), then two
Scale factor between a image is eζ', angle difference is θ '.By I1Become according to the scale factor and angle that calculate
It changes, the image after converting is denoted as I '1.According to I0With I '1, calculate peak valueFormula is as follows:
The position of peak value is I after converting0With I '1Between displacement.
(8) if in I1、I2、I2And I4Qualified matching is not found, then allows I0Transformation ginseng has been obtained with previous frame
Several images are matched, and matching process is the same as (4)~(7).Current position image is mixed with characteristic image, obtains having both two
The new images of characteristics of image are opened as new known location image;Pattern image finally is added in present frame, I1、I2、I2And I4
The middle pattern image nearest apart from present frame is deleted, and the update in pattern image library is carried out, to carry out subsequent matching work.
(9) the hash query table for establishing objects in images size and distance in advance, obtains in present frame according to scale factor
The size of object, table look-up to obtain present frame object to mechanical arm camera distance.
The present invention provides a kind of mobile robot cargo localization method based on Fourier-Mellin transform, with ground or goods
Image on frame demarcates shelf, then by Fourier-Mellin transform cargo image is calculated, it can be achieved that a variety of shelf and
The high accuracy positioning of cargo, while while passing through calibration point every time will be updated the image information of calibration, reach adaptively surface wear,
Situations such as variation, does not need to carry out cargo additional processing, applied widely and facilitate operation.
The basic principles, main features and advantages of the invention have been shown and described above.The technical staff of the industry should
Understand, the above embodiments do not limit the invention in any form, all obtained by the way of equivalent substitution or equivalent transformation
Technical solution is fallen within the scope of protection of the present invention.
Claims (8)
1. a kind of cargo localization method based on Fourier-Mellin transform, which comprises the steps of:
Step 1 establishes absolute coordinate system,
Before moveable robot movement,
Multiple calibration points are artificially set in coordinate system, and calibration point is arranged in for marking the position of shelf in movement routine, often
The absolute coordinate of a calibration point be it is known that positional relationship between mark point and shelf it is also known that;
It identifies calibration point, positions the location information of shelf;
Step 2 is protected from light place for what camera was mounted on that mobile robot takes object structure;
Step 3 establishes characteristics of image library,
Cargo on shelf is shot, characteristic pattern of the particular frame image as corresponding cargo location in extract real-time video
Picture, and then form a series of characteristic image of cargo on shelf;
The position orientation relation that current robot and shelf are obtained by label point image, calculates affine matrix, the spy that will be obtained before
Sign image obtains new characteristic image by affine transformation, is stored in cargo characteristic image library;
Step 4 obtains the cargo image for currently taking object locations of structures to shoot when taking object structure to start to position cargo,
If current signature image texture and the non-overlapping place in characteristic image library, are based on the image of previous frame known location
The image registration of Fourier-Mellin transform, currently to be taken the relative pose of object structure and cargo;
If current signature image texture has overlapping with characteristic image library, the two is carried out based on Fourier-Mellin transform
Image registration, to obtain the relative pose of current cargo;
Current location characteristic image is registrated, after the completion of registration, after affine transformation by step 5 with characteristic image library
Current position image is mixed with characteristic image, and the new images for obtaining having both two characteristics of image are as new known location figure
Picture deletes the pattern image nearest apart from present frame, carries out the update in pattern image library, and it is subsequent to jump to step 4 progress
Cargo position work.
2. a kind of cargo localization method based on Fourier-Mellin transform according to claim 1, which is characterized in that
In step 1, identify calibration point, position the location information of shelf specific method include: calibration point be fixed on ground or
On person's shelf, the relative position of calibration point and shelf be it is fixed and known, mobile robot clicks through calibration by camera
Row imaging, identification calibration point determine the number of shelf.
3. a kind of cargo localization method based on Fourier-Mellin transform according to claim 2, which is characterized in that
In step 1, the representation of the calibration point includes: one-dimension code, two dimensional code, additional character or special texture.
4. a kind of cargo localization method based on Fourier-Mellin transform according to claim 1, which is characterized in that
In step 2, place is protected from light by what camera was mounted on that mobile robot takes object structure;If the camera can not cover institute
There is range, then shelf are divided into multiple subregions, each subregion has movement of the mark point for robot.
5. a kind of cargo localization method based on Fourier-Mellin transform according to claim 1, which is characterized in that
In step 3, characteristics of image library is established,
Cargo is placed on shelf, the cargo on shelf is shot, the particular frame image in extract real-time video is as phase
The characteristic image of cargo location is answered, and then forms a series of characteristic image of cargo on shelf;Particular frame be extracted foundation be,
First frame is key frame, and the picture frame when acquisition time with previous frame is more than regular hour T is key frame, with previous frame
Picture frame when acquisition position is more than certain range M is key frame;
By the identification to mark point, the location information of shelf is obtained, affine transformation matrix is calculated, utilizes affine transformation matrix pair
The image of the shot by camera of Qu Wu mechanism carries out affine transformation and obtains new characteristic image, is stored in cargo characteristic image library;
The location information of the shelf mainly includes the relative position and relative angle between mobile robot and shelf;
The image of shooting need to cover the whole region of current shelf, if the coverage area of Qu Wu mechanism can not cover shelf whole area
Multiple calibration points are placed in domain before shelf, and shelf are divided into multiple subregions, cooperate the mobile realization of robot to shelf area
Covering.
6. a kind of cargo localization method based on Fourier-Mellin transform according to claim 1, which is characterized in that
In step 4, when taking object structure to start to position cargo, the cargo image for currently taking object locations of structures to shoot is obtained,
If current signature image texture and the non-overlapping place in characteristic image library, are based on the image of previous frame known location
The image registration of Fourier-Mellin transform, the relative pose with cargo currently to be taken object structure;
If current signature image texture have with characteristic image library it is overlapping, to current signature image and the characteristic image library for having overlapping
Image carries out the image registration based on Fourier-Mellin transform, to obtain the relative pose of current cargo;
The method for obtaining transverse shifting relationship between cargo and load-engaging device includes: the image arrived using current shooting and known bits
The cross energy spectrum for the image Fourier variation set, solves the translation relation of image, and then releases cargo and take the position of object structure
Relationship;Assuming that known image and the Fourier spectrum of present image are F1And F2, the cross energy spectrum of two images in frequency domain is calculated, it is public
Formula is as follows:
Cross energy is composed and carries out inverse transformation, peak value is obtained and is located at (x0,y0) impulse function, (x0,y0) be present image with
The position deviation for knowing image, by taking object structure initial position and camera internal reference, the transverse shifting for calculating cargo and Qu Wu mechanism is closed
System.
7. a kind of cargo localization method based on Fourier-Mellin transform according to claim 1, which is characterized in that
In step 4, when taking object structure to start to position cargo, the cargo image for currently taking object locations of structures to shoot is obtained,
If current signature image texture and the non-overlapping place in characteristic image library are carried out with the image of previous frame known location based on Fu
In leaf-Mellin transform image registration, currently to be taken the rotation relationship of object structure;
If current signature image texture has overlapping with characteristic image library, the two is carried out based on Fourier-Mellin transform
Image registration, to obtain the rotation relationship of current cargo;
The method for obtaining rotation angle between cargo and load-engaging device includes: the image for arriving current shooting and the figure of known location
As the energy conversion of Fourier spectrum is to polar coordinate system, the rotation relationship of image is solved, and then release cargo and take the rotation of object structure
Transfer the registration of Party membership, etc. from one unit to another;
Assuming that known image and the Fourier spectrum of present image are F1And F2, the corresponding energy spectrum of Fourier is calculated, formula is as follows:
Energy is converted to polar coordinate system M1(ρ, θ) and M2(ρ, θ), to M under polar coordinates1And M2It is related to solve phase, formula is as follows:
It obtains being located at θ0The angle of impulse function at position, i.e. present image and known image is θ0;
By taking object structure initial position and camera internal reference, the positional relationship of cargo and Qu Wu mechanism is calculated.
8. a kind of cargo localization method based on Fourier-Mellin transform according to claim 1, which is characterized in that
In step 4, when taking object structure to start to position cargo, the cargo image for currently taking object locations of structures to shoot is obtained,
If current signature image texture and the non-overlapping place in characteristic image library are carried out with the image of previous frame known location based on Fu
In leaf-Mellin transform image registration, currently to be taken the change of scale relationship of object structure;
If current signature image texture has overlapping with characteristic image library, the two is carried out based on Fourier-Mellin transform
Image registration, to obtain the change of scale relationship of current cargo;
The method for obtaining distance between cargo and load-engaging device includes: image Fu of the image for arriving current shooting and known location
In the energy conversion composed of leaf solve the change of scale relationship relationship of image to logarithmic coordinates system, and then be implemented in combination with the Kazakhstan of foundation
Uncommon table inquires distance representated by current scale, and then obtains cargo and take at a distance from object structure;Assuming that known image with work as
The Fourier spectrum of preceding image is F1(ξ, η) and F2They are converted into logarithmic coordinates system, F by (ξ, η)1(log ξ, log η) and F2
(log ξ, log η) it is related to solve phase:
The scale factor s of present image and known image is obtained, the hash query table of scale factor and actual distance is pre-established,
Can be obtained under current state cargo with take object it is inter-agency at a distance from.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910345255.4A CN110097494B (en) | 2019-04-26 | 2019-04-26 | Fourier-Mellin transform-based cargo positioning method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910345255.4A CN110097494B (en) | 2019-04-26 | 2019-04-26 | Fourier-Mellin transform-based cargo positioning method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110097494A true CN110097494A (en) | 2019-08-06 |
CN110097494B CN110097494B (en) | 2023-01-13 |
Family
ID=67446076
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910345255.4A Active CN110097494B (en) | 2019-04-26 | 2019-04-26 | Fourier-Mellin transform-based cargo positioning method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110097494B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111951318A (en) * | 2020-08-10 | 2020-11-17 | 上海科技大学 | Extended Fourier Mellin positioning algorithm applied to multi-depth scene |
CN115086539A (en) * | 2021-03-15 | 2022-09-20 | 虫极科技(北京)有限公司 | Shooting point positioning method and system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040218815A1 (en) * | 2003-02-05 | 2004-11-04 | Sony Corporation | Image matching system and image matching method and program |
CN106996777A (en) * | 2017-04-21 | 2017-08-01 | 合肥井松自动化科技有限公司 | A kind of vision navigation method based on ground image texture |
CN108373001A (en) * | 2018-02-01 | 2018-08-07 | 王学斌 | A kind of intelligent industrial robot automation warehousing system |
-
2019
- 2019-04-26 CN CN201910345255.4A patent/CN110097494B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040218815A1 (en) * | 2003-02-05 | 2004-11-04 | Sony Corporation | Image matching system and image matching method and program |
CN106996777A (en) * | 2017-04-21 | 2017-08-01 | 合肥井松自动化科技有限公司 | A kind of vision navigation method based on ground image texture |
CN108373001A (en) * | 2018-02-01 | 2018-08-07 | 王学斌 | A kind of intelligent industrial robot automation warehousing system |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111951318A (en) * | 2020-08-10 | 2020-11-17 | 上海科技大学 | Extended Fourier Mellin positioning algorithm applied to multi-depth scene |
CN111951318B (en) * | 2020-08-10 | 2023-08-04 | 上海科技大学 | Extended Fourier Merlin positioning algorithm applied to multi-depth scene |
CN115086539A (en) * | 2021-03-15 | 2022-09-20 | 虫极科技(北京)有限公司 | Shooting point positioning method and system |
CN115086539B (en) * | 2021-03-15 | 2024-02-02 | 虫极科技(北京)有限公司 | Shooting point positioning method and system |
Also Published As
Publication number | Publication date |
---|---|
CN110097494B (en) | 2023-01-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106651942B (en) | Three-dimensional rotating detection and rotary shaft localization method based on characteristic point | |
CN104778690B (en) | A kind of multi-target orientation method based on camera network | |
CN105234943B (en) | A kind of industrial robot teaching device and method of view-based access control model identification | |
CN104484648B (en) | Robot variable visual angle obstacle detection method based on outline identification | |
CN110378931A (en) | A kind of pedestrian target motion track acquisition methods and system based on multi-cam | |
CN109102547A (en) | Robot based on object identification deep learning model grabs position and orientation estimation method | |
CN110211180A (en) | A kind of autonomous grasping means of mechanical arm based on deep learning | |
CN104517289B (en) | A kind of indoor scene localization method based on hybrid camera | |
CN107038448A (en) | Target detection model building method | |
CN111563878B (en) | Space target positioning method | |
CN111242991B (en) | Method for quickly registering visible light and infrared camera | |
CN109544612A (en) | Point cloud registration method based on the description of characteristic point geometric jacquard patterning unit surface | |
CN112132874B (en) | Calibration-plate-free heterogeneous image registration method and device, electronic equipment and storage medium | |
CN109308722A (en) | A kind of spatial pose measuring system and method based on active vision | |
CN106096621B (en) | Based on vector constraint drop position detection random character point choosing method | |
CN105279769A (en) | Hierarchical particle filtering tracking method combined with multiple features | |
CN110097494A (en) | A kind of cargo localization method based on Fourier-Mellin transform | |
CN113470090A (en) | Multi-solid-state laser radar external reference calibration method based on SIFT-SHOT characteristics | |
CN112070831B (en) | Active camera repositioning method based on multi-plane joint pose estimation | |
CN106295512A (en) | Many correction line indoor vision data base construction method based on mark and indoor orientation method | |
CN108415434A (en) | A kind of robot dispatching method | |
CN109870106A (en) | A kind of building volume measurement method based on unmanned plane picture | |
CN109670501A (en) | Object identification and crawl position detection method based on depth convolutional neural networks | |
CN110532865A (en) | Spacecraft structure recognition methods based on visible light and laser fusion | |
CN113781525A (en) | Three-dimensional target tracking algorithm research based on original CAD model |
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 |