CN107123146A - The mark localization method and system of a kind of scaling board image - Google Patents
The mark localization method and system of a kind of scaling board image Download PDFInfo
- Publication number
- CN107123146A CN107123146A CN201710165758.4A CN201710165758A CN107123146A CN 107123146 A CN107123146 A CN 107123146A CN 201710165758 A CN201710165758 A CN 201710165758A CN 107123146 A CN107123146 A CN 107123146A
- Authority
- CN
- China
- Prior art keywords
- image
- scaling board
- mark
- interest
- profile
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
-
- 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
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of mark localization method of scaling board image and system, wherein, methods described to scaling board image by carrying out image pyramid layering;Binaryzation is proceeded by from top image, scaling board region and mark profile are searched in bianry image, if being continued search for if not searching in next tomographic image, to the last one layer or untill searching scaling board region and mark profile.The outsourcing rectangle of the mark profile in scaling board region is calculated in the pyramid diagram picture for search scaling board region, the outsourcing rectangle is mapped in the scaling board image, the centre coordinate of the area-of-interest comprising mark and calculation flag thing is obtained.It follows that beginning through Contour searching from the low-resolution image of top layer to find scaling board region and mark profile, the area-of-interest that can fast and accurately navigate to where the mark in image improves location efficiency.
Description
Technical field
The present invention relates to technical field of machine vision, and in particular to the mark localization method and system of scaling board image.
Background technology
In industrial 4.0 epoch, accurate measurement and detection based on machine vision are a highly important research fields, no
Pipe is that have this very wide application prospect in traditional manufacture or 3C electronics industry or aerospace field etc..Camera mark
Surely it is the basis that machine vision obtains accurate measurement data, camera calibration is by between world coordinate system and image coordinate system
Corresponding relation solves camera internal parameter and the process of external parameter, and the accuracy of actual measurement data is dependent in camera
The accuracy of portion's parameter and external parameter, then after all corresponding between one group of accurate world coordinate system and image coordinate system
Relation is the accurate primary factor of camera calibration.
The scaling board image that usual precision machined plane target drone and camera are gathered be used to determine two coordinate systems it
Between corresponding relation, due to the target based on circle marker thing relative to based on tessellated target have more preferable fault-tolerance with
More superior applicability, so prior art uses circle marker thing target, i.e., sets the M of equidistantly distributed on plane target drone
Individual filled circles.The actual physics coordinate (world coordinate system, unit is mm) in the circle marker thing center of circle is artificially to set on plane target drone
It is fixed, ensure precision by Precision Machining, and the pixel coordinate in the scaling board Circle in Digital Images shape mark center of circle of camera collection
(image coordinate system, unit is pixel) is obtained by image processing algorithm, and acquisition precision also depends primarily on mark and determined
The accuracy of position algorithm.
Current existing mark location algorithm mostly carries out the edge extracting or wheel of sub-pix to scaling board image artwork
Exterior feature is extracted, and carries out ellipse fitting to extract center of circle image coordinate, although can accurately be extracted central coordinate of circle, but be still suffered from following
Some is not enough:Scaling board image artwork is operated, efficiency of algorithm is relatively low when for high-resolution scaling board image, run
Time is long;When carrying out sub-pixel edge or contours extract, parameter setting is more, and tuning parameter repeatedly is needed during demarcation, for debugging
Personnel specialty requested knowledge is higher;It is higher to the image quality requirements of scaling board image, it is very quick to noise and illumination effect etc.
Sense, robustness is not strong.
Therefore, prior art has much room for improvement and improved.
The content of the invention
The application provides the mark localization method and system of a kind of scaling board image, to lift the effect of mark positioning
Rate.
According to the first aspect of the invention, the present invention provides a kind of mark localization method of scaling board image, including such as
Lower step:
Image layered step, the golden word of image is carried out to the scaling board image according to the resolution ratio of the scaling board image collected
Tower is layered, to obtain the pyramid diagram picture of each layer;
Step is performed, Contour searching step is performed to lower floor successively since the pyramid diagram picture of top layer, until current
Layer pyramid diagram is as untill searching scaling board region and mark profile in corresponding bianry image;When in current layer pyramid
When scaling board region and mark profile are searched in the corresponding bianry image of image, positioning step is performed;
Contour searching step, carries out image binaryzation processing to current layer pyramid diagram picture, obtains current layer pyramid diagram
As corresponding bianry image, scaling board region and mark profile are searched in the bianry image;
Positioning step, calculates the mark profile in scaling board region in the pyramid diagram picture for search scaling board region
Outsourcing rectangle, the outsourcing rectangle is mapped in the scaling board image, obtain comprising mark area-of-interest simultaneously
The centre coordinate of calculation flag thing.
The mark localization method of described scaling board image, wherein, the execution step also includes:
When not searching scaling board region and mark profile yet in the corresponding bianry image of pyramid bottom image,
The positioning failure of determination flag thing.
The mark localization method of described scaling board image, wherein, the Contour searching step, including:
Automatic two-value subslep, uses automatic threshold segmentation method to carry out image two-value on the pyramid diagram picture of current layer
Change is handled, and obtains bianry image;
Searching sub-step, Contour searching is carried out to obtained bianry image;
Judge sub-step, judge whether the outline searched includes the Internal periphery of predetermined number, the chi of the Internal periphery
It is very little whether all in the range of pre-set dimension;If so, then judging region that the outline included as scaling board region, Internal periphery
For mark profile;If it is not, then entering progressively two-value subslep;
Progressively two-value subslep, in predetermined threshold value is interval, is spaced predetermined interval value to choose gray scale threshold from small to large
Value, often chooses a gray threshold, then enters row threshold division to the pyramid diagram picture of current layer with the gray threshold, obtains correspondence
Bianry image, and searching sub-step is returned to, untill traversal predetermined threshold value is interval.
The mark localization method of described scaling board image, wherein, the searching sub-step, including:
Body corrosion treatment is done to obtained bianry image, and Contour searching is carried out to the bianry image after corrosion;Its
In, depending on the level of the template size size of corrosion according to pyramid diagram picture.
The mark localization method of described scaling board image, wherein, the positioning step, including:
The outsourcing of all mark profiles in scaling board region is calculated in the pyramid diagram picture for search scaling board region
Rectangle, the outsourcing rectangle is mapped in the scaling board image, is obtained including the area-of-interest of mark, is each felt emerging
Interesting region correspondence includes a mark;
Pretreatment is filtered using the filtering method at holding edge to each area-of-interest;
Image binaryzation processing is carried out using automatic threshold segmentation method to each area-of-interest, and calculated at binaryzation
The barycenter of the mark in area-of-interest after reason, a certain summit of the barycenter navigated to and area-of-interest is overlapped
Processing can obtain position coordinates of the center of mark in the scaling board image.
According to the second aspect of the invention, the present invention provides a kind of mark alignment system of scaling board image, wherein, bag
Include:
Image layered module, image is carried out for the resolution ratio according to the scaling board image collected to the scaling board image
Pyramid technology, to obtain the pyramid diagram picture of each layer;
Performing module, for being handled successively the pyramid diagram picture of top layer to bottom by Contour searching module, directly
Scaling board region and mark profile are searched in current layer pyramid diagram is as corresponding bianry image to Contour searching module
Untill;When Contour searching module in current layer pyramid diagram as corresponding bianry image in search scaling board region and mark
During profile, mark is positioned by locating module;
Contour searching module, for carrying out image binaryzation processing to current layer pyramid diagram picture, obtains the golden word of current layer
The corresponding bianry image of tower image, searches for scaling board region and mark profile in the bianry image;
Locating module, for calculating the mark in scaling board region in the pyramid diagram picture for search scaling board region
The outsourcing rectangle of profile, the outsourcing rectangle is mapped in the scaling board image, obtains including the region of interest of mark
Domain and the centre coordinate of calculation flag thing.
The mark alignment system of described scaling board image, wherein, the performing module is additionally operable to:
Contour searching module do not searched yet in the corresponding bianry image of the pyramid bottom image scaling board region and
During mark profile, the positioning failure of determination flag thing.
The mark alignment system of described scaling board image, wherein, the Contour searching module includes:
Automatic binarization unit, image two is carried out for the pyramid diagram picture to current layer using automatic threshold segmentation method
Value is handled, and obtains bianry image;
Progressively binarization unit, the outline for being searched in search unit does not include the Internal periphery of predetermined number, institute
When stating the size of Internal periphery not in the range of pre-set dimension, in predetermined threshold value is interval, predetermined interval value is spaced from small to large and is come
Gray threshold is chosen, a gray threshold is often chosen, then carries out threshold value point to the pyramid diagram picture of current layer with the gray threshold
Cut, obtain corresponding bianry image, untill traversal predetermined threshold value is interval;
Search unit, the bianry image for being obtained to automatic binarization unit carries out Contour searching;It is outer what is searched
Profile does not include the Internal periphery of predetermined number, when the size of the Internal periphery is not in the range of pre-set dimension, successively to progressively two
The bianry image that value unit is obtained carries out Contour searching;
Judging unit, it is described for judging whether outline that search unit is searched includes the Internal periphery of predetermined number
Whether the size of Internal periphery is all in the range of pre-set dimension;If so, then judging region that the outline included as scaling board
Region, Internal periphery is mark profile;If it is not, then starting progressively binarization unit.
The mark alignment system of described scaling board image, wherein, the search unit carries out profile in bianry image
Before search, body corrosion treatment is first done to bianry image;Wherein, the template size size of corrosion according to pyramid diagram as
Depending on level.
The mark alignment system of described scaling board image, wherein, the locating module, including:
Area-of-interest acquiring unit, for calculating scaling board region in the pyramid diagram picture for search scaling board region
The outsourcing rectangle of interior all mark profiles, the outsourcing rectangle is mapped in the scaling board image, is obtained comprising mark
The area-of-interest of thing, each area-of-interest correspondence includes a mark;
Filter unit, for being filtered pretreatment using the filtering method at holding edge to each area-of-interest;
Coordinate calculating unit, for being carried out to each area-of-interest using automatic threshold segmentation method at image binaryzation
Reason, and the barycenter of the mark in the area-of-interest after binary conversion treatment is calculated, by the barycenter navigated to and area-of-interest
A certain summit be overlapped processing and can obtain the position coordinates of the center of mark in the scaling board image.
Beneficial effects of the present invention:Image pyramid point is carried out to scaling board image using the down-sampled technology of image pyramid
Layer, begins through Contour searching to find scaling board region and mark profile from the low-resolution image of top layer, can be quick
Accurately the area-of-interest that navigates to where the mark in image, improves location efficiency, reduces debugging difficulty and not
Influence positioning precision., can not only by being filtered pretreatment using the filtering method at holding edge to each area-of-interest
Enough ensure the positioning precision of mark center, and improve the robustness of the reply phenomenon such as noise and uneven illumination.
Brief description of the drawings
In one embodiment of the mark alignment system for the scaling board image that Fig. 1 provides for the present invention, scaling board image
Schematic diagram;
The structured flowchart of one embodiment of the mark alignment system for the scaling board image that Fig. 2 provides for the present invention;
In one embodiment of the mark alignment system for the scaling board image that Fig. 3 provides for the present invention, Contour searching module
Structured flowchart;
The flow chart of the mark localization method for the scaling board image that Fig. 4 provides for the present invention;
The particular flow sheet of the mark localization method for the scaling board image that Fig. 5 provides for the present invention.
Embodiment
The present invention is described in further detail below by embodiment combination accompanying drawing.
The term definition used in the present invention:
Image pyramid is a kind of structure for carrying out interpretation of images with multiresolution.The pyramid of piece image be it is a series of with
The image collection that the resolution ratio of Pyramid arrangement is gradually reduced.Pyramidal bottom diagram seems high-resolution expression, and
Top is the expression of low resolution.
In area-of-interest (ROI, region of interest), machine vision, image procossing, from processed image
Being sketched the contours of with modes such as square frame, circle, ellipse, irregular polygons needs to use rectangular box in region to be processed, the present embodiment.
The present invention carries out the image pyramid needed for calculating determination to the resolution dimensions of scaling board image (target image)
The number of plies;It is down-sampled according to required number of plies progress image pyramid, and preserve each tomographic image;Two-value is proceeded by from top image
Change, Morphological scale-space and determines whether to search target (scaling board) region and mark profile Contour searching, such as judge failure
Then next tomographic image is scanned for judge, untill finding, all number of plies images has such as been searched for and has not all found scaling board area
Domain, then determination flag thing positioning failure;Scaling board region is such as found, then extracts the profile of mark in scaling board region, and is counted
Its boundary rectangle is calculated, four summits of each boundary rectangle are mapped in the scaling board image by pyramid, will be newly outer
Rectangle is connect as area-of-interest (ROI);Noise and uneven illumination are eliminated using the filtering method at holding edge to each ROI
Influence;Automatic threshold segmentation is carried out to the circle marker in each ROI to ask for barycenter to be accurately positioned the centre coordinate of mark.
Illustrated below by specific embodiment.In embodiment one, Fig. 1 and Fig. 2 are refer to, the present invention provides one kind
The mark alignment system of scaling board image, including:Image layered module 10, performing module 20, Contour searching module 30 and fixed
Position module 40.
Image layered module 10, for gathering scaling board image (as shown in Figure 1), according to the scaling board image collected
Resolution ratio carries out image pyramid layering to the scaling board image, to obtain the pyramid diagram picture of each layer.Gather scaling board image
It can be direct shooting scaling board image, the extraneous scaling board image transmitted can also be received.Gather after scaling board image,
The down-sampled number of plies of the image pyramid according to needed for being calculated the resolution ratio of scaling board image.Calculation criterion is:Image pyramid drops
The line number and/or columns of the top layer images of sampling cannot be below default minimum limitation line number Rmin.Counted since scaling board image
Calculate, the line number and/or columns of current tomographic image are halved, the line number and/or columns of last layer image is obtained, until line number and/
Or columns is less than default minimum limitation line number RminUntill.Thus calculate and obtain pyramidal the number of plies N+1 layer, N for more than or equal to
1 integer.It is the 0th layer of pyramid, the i.e. bottom to provide scaling board image.
Image layered module 10 is additionally operable to, and the down-sampled n times of pyramid is carried out to the scaling board image, and every layer is sampled
Image preserve.Specifically, the resolution ratio of scaling board image is reduced into original a quarter, (line number and columns reduce one
Half) last layer (the 1st layer) pyramid diagram picture is obtained, then the resolution ratio of the obtain the 1st layer of pyramid diagram picture is reduced to original
A quarter, obtains the 2nd layer of pyramid diagram picture, the rest may be inferred, until the line number and columns of pyramid diagram picture are less than RminUntill.
That is, the line number and columns of the pyramid diagram picture of top layer (N+1 layers) are more than Rmin, its resolution ratio halve after image line number and
Columns is less than Rmin。RminSpan be empirical value, specific scope is 65-127, in the present embodiment, RminValue is 100.
In practice, in order to avoid the visual field wastes and demarcated accurate, scaling board region has to account for entire image big
Small more than 1/4, and pyramid is down-sampled that image can be all reduced to original 1/4, such as scaling board image resolution ratio every time
For 2048*2048, the 1st time it is down-sampled after be 1024*1024, the 2nd time it is down-sampled after be 512*512, the 3rd time it is down-sampled after be
256*256, is 128*128 after the 4th is down-sampled, and when progress the 5th is down-sampled, image resolution ratio is reduced to 64*64, now
The 5th layer of pyramid diagram picture it is too small so that it cannot find the profile of the outline and mark of scaling board.Therefore R is setmin's
Scope can reduce unnecessary layering, and its specific scope, which is arranged on 65-127, can ensure the accuracy of follow-up Contour searching.
Performing module 20, for being located successively to the pyramid diagram picture of top layer to bottom by Contour searching module 30
Reason, until Contour searching module 30 searches scaling board region and mark in current layer pyramid diagram is as corresponding bianry image
Untill thing profile;When Contour searching module 30 in current layer pyramid diagram as corresponding bianry image in search scaling board region
And during mark profile, mark is positioned by locating module 40;Contour searching module 30 is in pyramid bottom figure
As not searched yet in corresponding bianry image when scaling board region and mark profile, the positioning failure of determination flag thing.Due to
The pixel of the pyramid diagram picture of top layer is minimum, and scaling board region and mark profile are searched for since the image of lowest resolution,
The area-of-interest that can fast and accurately navigate to where the mark in image, improves location efficiency.
Mark can be that circle marker thing is used in circular, oval or other shapes of mark, the present embodiment, such as
Shown in Fig. 1.
Contour searching module 30, for carrying out image binaryzation processing to current layer pyramid diagram picture, obtains current layer gold
The corresponding bianry image of word tower image, searches for scaling board region and mark profile in the bianry image.Specifically, referring to
Fig. 3, Contour searching module 30 includes:Automatic binarization unit 310, progressively binarization unit 340, search unit 320 and judgement
Unit 330.
Automatic binarization unit 310, is schemed for the pyramid diagram picture to current layer using automatic threshold segmentation method
As binary conversion treatment, bianry image is obtained.Automatic threshold segmentation method is included in Otsu methods, maximum entropy method (MEM) and Adaptive Thresholding
One or more.
Progressively binarization unit 340, the outline for being searched in search unit 320 does not include the lubrication groove of predetermined number
Exterior feature, when the size of the Internal periphery is not in the range of pre-set dimension, in predetermined threshold value interval (TminTo Tmax) in, from small to large between
Gray threshold is chosen every predetermined interval value △ T, a gray threshold is often chosen, then with golden word of the gray threshold to current layer
Tower image enters row threshold division, obtains corresponding bianry image, untill traversal predetermined threshold value is interval.Predetermined threshold value it is interval and
Predetermined interval value △ T can be configured as needed.
Search unit 320, for first doing body corrosion treatment to the bianry image that automatic binarization unit 310 is obtained,
Contour searching is carried out again;The Internal periphery of predetermined number is not included in the outline searched, the size of the Internal periphery is not pre-
If when in size range, first doing body corrosion treatment to the bianry image that progressively binarization unit 340 is obtained successively, then carry out
Contour searching.Wherein, depending on level of the template size size of corrosion according to pyramid diagram picture.The method of Contour searching can be used
Chain-Code-Method or Run- Length Coding method etc..
Judging unit 330, for judging whether outline that search unit 320 is searched includes the lubrication groove of predetermined number
Exterior feature, whether the size of the Internal periphery is all in the range of pre-set dimension;If so, then judge region that the outline included as
Scaling board region, Internal periphery is mark profile;If it is not, then starting progressively binarization unit 340.Search unit 320 is searched for
To outline be rectangle frame in scaling board image inner edge (rectangle frame has inner edge and outside), without consider its outside, it is only right
Inner edge makes a decision.Internal periphery be rectangle frame in mark edge.Therefore, predetermined number is actually to be wrapped on scaling board image
The number of filled circles in the quantity of the mark contained, i.e. Fig. 1.Pre-set dimension scope is that can filter out the size model of mark
Enclose.Size includes area and girth.That is, whether the size of Internal periphery is all in the range of pre-set dimension, specially:The face of Internal periphery
Whether product and girth are respectively in preset area scope, default peripheral extent.Circle, ellipse, rectangle, square etc., in radius
Or in the case of the length of side is known, its area and girth by limiting the size of area and girth just, it is known that so as to be told
Corresponding shape.
The present invention is using body corrosion, and effect is to expand outwardly the profile of the circle marker thing of black, is easy to follow-up
After the profile of circle marker thing is extracted, when calculating the outsourcing rectangle of this profile, it is ensured that this rectangle can be by the scaling board
The circle marker thing of image is included, and one piece of situation is intersected or be connected on the side that rectangle is not present with circle marker thing profile,
Ensure that positioning precision.
Locating module 40, for calculating the mark in scaling board region in the pyramid diagram picture for search scaling board region
The outsourcing rectangle of thing profile, the outsourcing rectangle is mapped in the scaling board image, obtains interested comprising mark
Region and the centre coordinate of calculation flag thing.
Further, locating module 40, including:Area-of-interest acquiring unit, filter unit and coordinate calculating unit.
Area-of-interest acquiring unit, for calculating scaling board region in the pyramid diagram picture for search scaling board region
The outsourcing rectangle of interior all mark profiles, the scaling board is mapped back according to the down-sampled principle of pyramid by the outsourcing rectangle
In image, obtain including the area-of-interest (ROI) of mark, each area-of-interest correspondence includes a mark.
Filter unit, for being filtered pretreatment using the filtering method at holding edge to each area-of-interest.Example
Such as:Keep the filtering method at edge can be using anisotropic diffusion filtering or bilateral filtering etc..Keep the filtering method energy at edge
Enough while marginal information is kept, good smothing filtering effect is played to noise and non-edge, can effectively be suppressed
The phenomenon such as noise and uneven illumination.
Coordinate calculating unit, for being carried out to each area-of-interest using automatic threshold segmentation method at image binaryzation
Reason, and the barycenter of the mark in the area-of-interest after binary conversion treatment is calculated, by the barycenter navigated to and area-of-interest
A certain summit (the present embodiment is left upper apex) be overlapped processing and can obtain the center of mark in the scaling board figure
Position coordinates as in.Likewise, automatic threshold segmentation method can be used:In Otsu methods, maximum entropy method (MEM) and Adaptive Thresholding
One or more.Wherein, the barycenter includes gray scale barycenter and/or region barycenter.The present invention is using the sense for asking for binaryzation
The mode of interest region barycenter asks for the center of mark, it is ensured that accuracy.
In other embodiments, locating module 40 can also extract the centre coordinate of mark using other methods, e.g.,
The outsourcing rectangle of the mark profile calculated in the pyramid diagram picture in scaling board region in scaling board region is searched, will be described outer
Bag rectangle is mapped in the scaling board image, obtains including the area-of-interest of mark;Combined using Boundary extracting algorithm
Ellipses recognition algorithm is accurately positioned the centre coordinate of mark.Boundary extracting algorithm includes Canny edge detection algorithms, LOG
Edge detection algorithm and Facet model Sub-pixel Edge Detections etc..
In summary, the mark alignment system for the scaling board image that the present invention is provided, passes through image pyramid technology energy
Enough area-of-interests quickly navigated in the scaling board image, follow-up mark barycenter extraction algorithm is felt just for these
Interest region is operated, therefore greatly improves operational efficiency.By coupling and using algorithm parameter with the pyramid number of plies
The method of Automatic parameter selection avoids extra parameter input, greatly facilitates algorithm debugging process, reduces debugging difficult
Degree.Can be existing for picture noise and uneven illumination etc. in the case where ensureing required precision using holding edge filter method
As there is good inhibiting effect, can greatly boosting algorithm robustness.
The mark alignment system of the scaling board image provided based on above-described embodiment, the present invention also provides a kind of scaling board
The mark localization method of image, as shown in figure 4, methods described comprises the following steps:
S10, image layered step:Image layered module gathers scaling board image, according to the scaling board image collected
Resolution ratio carries out image pyramid layering to the scaling board image, to obtain the pyramid diagram picture of each layer.Please refer to fig. 5,
Image layered step S10 is specifically included:S110, input scaling board image.
S120, image pyramid layering carried out to scaling board image according to the resolution ratio of scaling board image, and every layer is adopted
The image of sample is preserved.That is, the down-sampled number of plies of the image pyramid according to needed for being calculated the resolution ratio of scaling board image, to the mark
Fixed board image carries out the down-sampled n times of pyramid, and the image of every layer of sampling is preserved.
S20, execution step (being made up of S210, S220, S230):Held successively to lower floor since the pyramid diagram picture of top layer
Road wheel exterior feature search step S30, until in current layer pyramid diagram as corresponding bianry image in search scaling board region and mark
Untill will thing profile;When in current layer pyramid diagram as corresponding bianry image in search scaling board region and mark profile
When, perform positioning step S40;Scaling board region and mark are not searched yet in the corresponding bianry image of pyramid bottom image
During will thing profile, the positioning failure of determination flag thing.
Step is performed to be made up of S210, S220, S230.Wherein, S210, choose i-th layer of pyramid diagram picture, selection it is initial
It is worth for top layer pyramid diagram picture, i.e. n-th layer pyramid diagram picture.I is more than or equal to the 0, integer less than or equal to N.In the present embodiment, i
More than or equal to 1.
S30, Contour searching step:Image binaryzation processing is carried out to current layer pyramid diagram picture, the golden word of current layer is obtained
The corresponding bianry image of tower image, searches for scaling board region and mark profile in the bianry image.
Specifically, Contour searching step S30 is specifically included:
S310, automatic threshold segmentation, i.e., automatic two-value subslep:Automatic binarization unit is to the pyramid currently chosen
Image carries out image binaryzation processing using automatic threshold segmentation method, obtains bianry image.
Searching sub-step, search unit carries out Contour searching to obtained bianry image.Specifically, searching sub-step bag
Include:S321, body corrosion treatment is done to obtained bianry image, wherein, the template size size of corrosion is according to pyramid diagram
Depending on the level of picture.S322, to after corrosion bianry image carry out Contour searching.
S330, judge sub-step:Judging unit judges whether the outline searched includes the Internal periphery of predetermined number, institute
The size of Internal periphery is stated whether all in the range of pre-set dimension;If so, then judging region that the outline included as demarcation
Plate region, Internal periphery is mark profile;If it is not, then entering progressively two-value subslep S340;
S340, threshold value self-adjusting and segmentation, i.e. progressively two-value subslep:Progressively binarization unit is interval in predetermined threshold value
It is interior, predetermined interval value is spaced from small to large to choose gray threshold, a gray threshold is often chosen, then with the gray threshold to working as
The pyramid diagram picture of front layer enters row threshold division, obtains corresponding bianry image, and returns to searching sub-step, until traversal is default
Untill threshold interval.Specifically, choosing the interval minimum value T of predetermined threshold valueminAs gray threshold, with the gray threshold to working as
The pyramid diagram picture of front layer enters row threshold division, obtains corresponding bianry image, and return to step S321;It is again introduced into step
During S340, last gray threshold increase predetermined interval value △ T are obtained into new gray threshold, with new gray threshold to working as
The pyramid diagram picture of front layer enters row threshold division, obtains corresponding bianry image, and return to step S321;Circulate according to this, until
Finally choose the interval maximum of T of predetermined threshold valuemaxAs gray threshold, with pyramid diagram picture of the gray threshold to current layer
Enter row threshold division, untill obtaining corresponding bianry image, and return to step S321.
The judged result of S230, performing module in S330, judges whether to search scaling board in current tomographic image
Region and mark profile, after S330 judges to obtain scaling board region and mark profile, then into step S410;In traversal
When not searching scaling board region and mark profile yet after predetermined threshold value is interval, into step S220.
S220, return i-1 value give step S210 so that performing module obtains the i-th -1 layer of pyramid diagram picture.
S40, positioning step:Locating module is calculated in scaling board region in the pyramid diagram picture for search scaling board region
Mark profile outsourcing rectangle, the outsourcing rectangle is mapped in the scaling board image, obtained comprising mark
The centre coordinate of area-of-interest and calculation flag thing.
Specifically, positioning step S40, including:
S410, area-of-interest acquiring unit calculate scaling board region in the pyramid diagram picture for search scaling board region
The outsourcing rectangle of interior all mark profiles, the outsourcing rectangle is mapped in the scaling board image, is obtained comprising mark
The area-of-interest of thing, each area-of-interest correspondence includes a mark.
S420, filter unit are filtered pretreatment to each area-of-interest using the filtering method at holding edge.
S430, coordinate calculating unit are carried out at image binaryzation to each area-of-interest using automatic threshold segmentation method
Reason, and the barycenter of the mark in the area-of-interest after binary conversion treatment is calculated, by the barycenter navigated to and area-of-interest
A certain summit (be specially left upper apex) be overlapped processing and can obtain the center of mark in the scaling board image
Position coordinates.
Due to the principle of the mark localization method of scaling board image, feature in above-mentioned corresponding embodiment of the method it is detailed
It is thin to illustrate, it will not be repeated here.
It will be understood by those skilled in the art that all or part of step of various methods can pass through in above-mentioned embodiment
Program controls related hardware to complete, and the program can be stored in a computer-readable recording medium, storage medium can be wrapped
Include:Read-only storage, random access memory, disk or CD etc..
Use above specific case is illustrated to the present invention, is only intended to help and is understood the present invention, not to limit
The system present invention.For those skilled in the art, according to the thought of the present invention, it can also make some simple
Deduce, deform or replace.
Claims (10)
1. a kind of mark localization method of scaling board image, it is characterised in that comprise the following steps:
Image layered step, image pyramid point is carried out according to the resolution ratio of the scaling board image collected to the scaling board image
Layer, to obtain the pyramid diagram picture of each layer;
Step is performed, Contour searching step is performed to lower floor successively since the pyramid diagram picture of top layer, until in current layer gold
Untill scaling board region and mark profile are searched in the corresponding bianry image of word tower image;When in current layer pyramid diagram picture
When scaling board region and mark profile are searched in corresponding bianry image, positioning step is performed;
Contour searching step, carries out image binaryzation processing to current layer pyramid diagram picture, obtains current layer pyramid diagram picture pair
The bianry image answered, searches for scaling board region and mark profile in the bianry image;
Positioning step, calculates the outer of mark profile in scaling board region in the pyramid diagram picture for search scaling board region
Bag rectangle, the outsourcing rectangle is mapped in the scaling board image, obtains the area-of-interest comprising mark and calculating
The centre coordinate of mark.
2. the mark localization method of scaling board image according to claim 1, it is characterised in that the execution step is also
Including:
When not searching scaling board region and mark profile yet in the corresponding bianry image of pyramid bottom image, judge
Mark positioning failure.
3. the mark localization method of scaling board image according to claim 1 or 2, it is characterised in that the profile is searched
Rope step, including:
Automatic two-value subslep, is carried out at image binaryzation to the pyramid diagram picture of current layer using automatic threshold segmentation method
Reason, obtains bianry image;
Searching sub-step, Contour searching is carried out to obtained bianry image;
Judge sub-step, judge whether the outline searched includes the Internal periphery of predetermined number, the size of the Internal periphery is
It is no all in the range of pre-set dimension;If so, then judging region that the outline included as scaling board region, Internal periphery is mark
Will thing profile;If it is not, then entering progressively two-value subslep;
Progressively two-value subslep, in predetermined threshold value is interval, is spaced predetermined interval value to choose gray threshold, often from small to large
A gray threshold is chosen, then row threshold division is entered to the pyramid diagram picture of current layer with the gray threshold, obtains corresponding two
It is worth image, and returns to searching sub-step, untill traversal predetermined threshold value is interval.
4. the mark localization method of scaling board image according to claim 3, it is characterised in that the search sub-step
Suddenly, including:
Body corrosion treatment is done to obtained bianry image, and Contour searching is carried out to the bianry image after corrosion;Wherein, it is rotten
Depending on level of the template size size of erosion according to pyramid diagram picture.
5. the mark localization method of scaling board image according to claim 1, it is characterised in that the positioning step,
Including:
The outsourcing rectangle of all mark profiles in scaling board region is calculated in the pyramid diagram picture for search scaling board region,
The outsourcing rectangle is mapped in the scaling board image, obtains including the area-of-interest of mark, each region of interest
Domain correspondence includes a mark;
Pretreatment is filtered using the filtering method at holding edge to each area-of-interest;
Image binaryzation processing is carried out using automatic threshold segmentation method to each area-of-interest, and calculated after binary conversion treatment
Area-of-interest in mark barycenter, a certain summit of the barycenter navigated to and area-of-interest is overlapped processing
It can obtain position coordinates of the center of mark in the scaling board image.
6. a kind of mark alignment system of scaling board image, it is characterised in that including:
Image layered module, the golden word of image is carried out to the scaling board image for the resolution ratio according to the scaling board image collected
Tower is layered, to obtain the pyramid diagram picture of each layer;
Performing module, for being handled successively the pyramid diagram picture of top layer to bottom by Contour searching module, until wheel
Wide search module during current layer pyramid diagram is as corresponding bianry image untill scaling board region and mark profile is searched;
When Contour searching module in current layer pyramid diagram as corresponding bianry image in search scaling board region and mark profile
When, mark is positioned by locating module;
Contour searching module, for carrying out image binaryzation processing to current layer pyramid diagram picture, obtains current layer pyramid diagram
As corresponding bianry image, scaling board region and mark profile are searched in the bianry image;
Locating module, for calculating the mark profile in scaling board region in the pyramid diagram picture for search scaling board region
Outsourcing rectangle, the outsourcing rectangle is mapped in the scaling board image, obtain comprising mark area-of-interest simultaneously
The centre coordinate of calculation flag thing.
7. the mark alignment system of scaling board image according to claim 6, it is characterised in that the performing module is also
For:
Contour searching module does not search scaling board region and mark yet in the corresponding bianry image of pyramid bottom image
During thing profile, the positioning failure of determination flag thing.
8. the mark alignment system of the scaling board image according to claim 6 or 7, it is characterised in that the profile is searched
Rope module includes:
Automatic binarization unit, image binaryzation is carried out for the pyramid diagram picture to current layer using automatic threshold segmentation method
Processing, obtains bianry image;
Progressively binarization unit, the outline for being searched in search unit does not include the Internal periphery of predetermined number, it is described in
When the size of profile is not in the range of pre-set dimension, in predetermined threshold value is interval, predetermined interval value is spaced from small to large to choose
Gray threshold, often chooses a gray threshold, then enters row threshold division to the pyramid diagram picture of current layer with the gray threshold, obtain
To corresponding bianry image, untill traversal predetermined threshold value is interval;
Search unit, the bianry image for being obtained to automatic binarization unit carries out Contour searching;In the outline searched
The Internal periphery of predetermined number is not included, when the size of the Internal periphery is not in the range of pre-set dimension, successively to progressively binaryzation
The bianry image that unit is obtained carries out Contour searching;
Judging unit, for judging whether outline that search unit is searched includes the Internal periphery of predetermined number, the lubrication groove
Whether wide size is all in the range of pre-set dimension;If so, region that the outline included then is judged as scaling board region,
Internal periphery is mark profile;If it is not, then starting progressively binarization unit.
9. the mark alignment system of scaling board image according to claim 8, it is characterised in that the search unit exists
Bianry image is carried out before Contour searching, and body corrosion treatment is first done to bianry image;Wherein, the template size size of corrosion
Depending on the level of pyramid diagram picture.
10. the mark alignment system of scaling board image according to claim 1, it is characterised in that the locating module,
Including:
Area-of-interest acquiring unit, for calculating institute in scaling board region in the pyramid diagram picture for search scaling board region
There is the outsourcing rectangle of mark profile, the outsourcing rectangle is mapped in the scaling board image, obtain comprising mark
Area-of-interest, each area-of-interest correspondence includes a mark;
Filter unit, for being filtered pretreatment using the filtering method at holding edge to each area-of-interest;
Coordinate calculating unit, for carrying out image binaryzation processing using automatic threshold segmentation method to each area-of-interest,
And the barycenter of the mark in the area-of-interest after binary conversion treatment is calculated, by certain of the barycenter navigated to and area-of-interest
One summit is overlapped processing and can obtain position coordinates of the center of mark in the scaling board image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710165758.4A CN107123146A (en) | 2017-03-20 | 2017-03-20 | The mark localization method and system of a kind of scaling board image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710165758.4A CN107123146A (en) | 2017-03-20 | 2017-03-20 | The mark localization method and system of a kind of scaling board image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107123146A true CN107123146A (en) | 2017-09-01 |
Family
ID=59718291
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710165758.4A Pending CN107123146A (en) | 2017-03-20 | 2017-03-20 | The mark localization method and system of a kind of scaling board image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107123146A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108280894A (en) * | 2017-12-08 | 2018-07-13 | 浙江国自机器人技术有限公司 | Not parking method for inspecting for power equipment |
CN108876860A (en) * | 2018-04-28 | 2018-11-23 | 中国计量大学 | A kind of image calibration method for pipe bubble offset measurement |
CN109509200A (en) * | 2018-12-26 | 2019-03-22 | 深圳市繁维医疗科技有限公司 | Checkerboard angle point detection process, device and computer readable storage medium based on contours extract |
CN110503605A (en) * | 2019-08-27 | 2019-11-26 | Oppo广东移动通信有限公司 | A kind of image processing method, device and storage medium |
CN110853104A (en) * | 2020-01-15 | 2020-02-28 | 广东博智林机器人有限公司 | Calibration plate, machine vision calibration device and method |
CN111161289A (en) * | 2019-12-26 | 2020-05-15 | 杭州格像科技有限公司 | Method, device and computer program product for improving contour precision of object in image |
CN112033408A (en) * | 2020-08-27 | 2020-12-04 | 河海大学 | Paper-pasted object space positioning system and positioning method |
CN112429444A (en) * | 2020-11-02 | 2021-03-02 | 武汉理工大学 | Garbage recycling robot and control method thereof |
CN113344955A (en) * | 2021-05-25 | 2021-09-03 | 云南凤云高速公路有限公司 | System and method for accurately detecting slope by fuzzy edge |
CN114529613A (en) * | 2021-12-15 | 2022-05-24 | 深圳市华汉伟业科技有限公司 | Method for extracting characteristic point high-precision coordinates of circular array calibration plate |
CN116894841A (en) * | 2023-09-08 | 2023-10-17 | 山东天鼎舟工业科技有限公司 | Visual detection method for quality of alloy shell of gearbox |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102346015A (en) * | 2011-06-24 | 2012-02-08 | 西安工程大学 | Method for measuring icing thickness of insulator of power transmission line based on video difference analysis |
CN104036244A (en) * | 2014-06-09 | 2014-09-10 | 西安邮电大学 | Checkerboard pattern corner point detecting method and device applicable to low-quality images |
-
2017
- 2017-03-20 CN CN201710165758.4A patent/CN107123146A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102346015A (en) * | 2011-06-24 | 2012-02-08 | 西安工程大学 | Method for measuring icing thickness of insulator of power transmission line based on video difference analysis |
CN104036244A (en) * | 2014-06-09 | 2014-09-10 | 西安邮电大学 | Checkerboard pattern corner point detecting method and device applicable to low-quality images |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108280894A (en) * | 2017-12-08 | 2018-07-13 | 浙江国自机器人技术有限公司 | Not parking method for inspecting for power equipment |
CN108876860A (en) * | 2018-04-28 | 2018-11-23 | 中国计量大学 | A kind of image calibration method for pipe bubble offset measurement |
CN108876860B (en) * | 2018-04-28 | 2023-06-20 | 中国计量大学 | Image calibration method for measuring bubble offset of tube level |
CN109509200A (en) * | 2018-12-26 | 2019-03-22 | 深圳市繁维医疗科技有限公司 | Checkerboard angle point detection process, device and computer readable storage medium based on contours extract |
CN109509200B (en) * | 2018-12-26 | 2023-09-29 | 深圳市繁维医疗科技有限公司 | Checkerboard corner detection method based on contour extraction and computer readable storage medium |
CN110503605B (en) * | 2019-08-27 | 2023-03-24 | Oppo广东移动通信有限公司 | Image processing method, device and storage medium |
CN110503605A (en) * | 2019-08-27 | 2019-11-26 | Oppo广东移动通信有限公司 | A kind of image processing method, device and storage medium |
CN111161289A (en) * | 2019-12-26 | 2020-05-15 | 杭州格像科技有限公司 | Method, device and computer program product for improving contour precision of object in image |
CN111161289B (en) * | 2019-12-26 | 2023-08-08 | 杭州格像科技有限公司 | Method and device for improving contour precision of object in image |
CN110853104A (en) * | 2020-01-15 | 2020-02-28 | 广东博智林机器人有限公司 | Calibration plate, machine vision calibration device and method |
CN110853104B (en) * | 2020-01-15 | 2020-05-05 | 广东博智林机器人有限公司 | Calibration plate, machine vision calibration device and method |
CN112033408B (en) * | 2020-08-27 | 2022-09-30 | 河海大学 | Paper-pasted object space positioning system and positioning method |
CN112033408A (en) * | 2020-08-27 | 2020-12-04 | 河海大学 | Paper-pasted object space positioning system and positioning method |
CN112429444A (en) * | 2020-11-02 | 2021-03-02 | 武汉理工大学 | Garbage recycling robot and control method thereof |
CN113344955A (en) * | 2021-05-25 | 2021-09-03 | 云南凤云高速公路有限公司 | System and method for accurately detecting slope by fuzzy edge |
CN114529613A (en) * | 2021-12-15 | 2022-05-24 | 深圳市华汉伟业科技有限公司 | Method for extracting characteristic point high-precision coordinates of circular array calibration plate |
CN114529613B (en) * | 2021-12-15 | 2024-07-09 | 深圳市华汉伟业科技有限公司 | Feature point high-precision coordinate extraction method for circular array calibration plate |
CN116894841A (en) * | 2023-09-08 | 2023-10-17 | 山东天鼎舟工业科技有限公司 | Visual detection method for quality of alloy shell of gearbox |
CN116894841B (en) * | 2023-09-08 | 2023-11-28 | 山东天鼎舟工业科技有限公司 | Visual detection method for quality of alloy shell of gearbox |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107123146A (en) | The mark localization method and system of a kind of scaling board image | |
CN109377485B (en) | Machine vision detection method for instant noodle packaging defects | |
CN104990925B (en) | One kind is based on gradient multi thresholds optimization defect inspection method | |
CN109636772A (en) | The defect inspection method on the irregular shape intermetallic composite coating surface based on deep learning | |
CN111027446B (en) | Coastline automatic extraction method of high-resolution image | |
CN105740945A (en) | People counting method based on video analysis | |
CN107451588A (en) | A kind of pop can smooth surface coding ONLINE RECOGNITION method based on machine vision | |
CN114219794B (en) | Method and system for evaluating surface quality of shaving board based on machine vision | |
CN107590512B (en) | The adaptive approach and system of parameter in a kind of template matching | |
CN109840483B (en) | Landslide crack detection and identification method and device | |
CN101984916A (en) | Blood vessel diameter measuring method based on digital image processing technology | |
CN107301408A (en) | Human body mask extracting method and device | |
CN101334896A (en) | Processing method for measuring sub-pixel rim of digital picture | |
CN109816051A (en) | A kind of harmful influence cargo characteristic point matching method and system | |
WO2023134251A1 (en) | Clustering-based light bar extraction method and apparatus | |
CN116777877A (en) | Circuit board defect detection method, device, computer equipment and storage medium | |
CN105913425B (en) | A kind of more pig contour extraction methods based on adaptive oval piecemeal and wavelet transformation | |
CN110097549A (en) | Based on morphologic land, water and air boundary line detecting method, system, medium and equipment | |
CN116563262A (en) | Building crack detection algorithm based on multiple modes | |
CN109544513A (en) | A kind of steel pipe end surface defect extraction knowledge method for distinguishing | |
CN109961065A (en) | A kind of surface vessel object detection method | |
CN107977976A (en) | A kind of image partition method | |
CN106570878A (en) | Heavy oil microcosmic interface detection method based on gray scale difference | |
CN117710638A (en) | Small target detection method based on single-stage anchor-free frame detection network | |
CN106372593B (en) | Optic disk area positioning method based on vascular convergence |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170901 |
|
RJ01 | Rejection of invention patent application after publication |