CN107452002A - A kind of image partition method and device - Google Patents
A kind of image partition method and device Download PDFInfo
- Publication number
- CN107452002A CN107452002A CN201610377503.XA CN201610377503A CN107452002A CN 107452002 A CN107452002 A CN 107452002A CN 201610377503 A CN201610377503 A CN 201610377503A CN 107452002 A CN107452002 A CN 107452002A
- Authority
- CN
- China
- Prior art keywords
- region
- candidate
- area
- candidate subimage
- subimage region
- 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
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a kind of image partition method and device.This method includes：It is gray level image by target image processing, and obtains the candidate subimage region included in the gray level image；The filtration treatment of setting strategy is carried out to the candidate subimage region included in the gray level image；The segmentation result to the target image is obtained according to filtration treatment result.The technical scheme of the embodiment of the present invention, by carrying out the filtration treatment of setting strategy to candidate subimage region, to filter out the impurity subgraph in candidate subimage region, thus reduce the subgraph that mistake is adulterated in target subgraph, improve and cut figure effect.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to a kind of image partition method and device.
Background technology
Image refers to dividing the image into several regions specific, with unique properties and proposes mesh interested
Target technology and process.Image segmentation is usually used in medical image analysis and the analysis of mechanical engineering image.
In existing image partition method, after carrying out gray processing processing to target image, generally carried out using upright projection
Coarse segmentation or progress edge detection process and Contour searching processing are split, to obtain target included in target image
Image.However, during inventor realizes the present invention, it is found that prior art has following defect：Vertically thrown based on gray processing
Profile processing, the impurity subgraph being all unable in Filtration Goal subgraph, so as to cause to split are searched for after shadow or rim detection
The subgraph of mistake is doped with obtained target subgraph, i.e. existing image partition method to cut figure ineffective.
The content of the invention
In view of this, the embodiment of the present invention provides a kind of image partition method and device, to improve image partition method
Cut figure effect.
In a first aspect, the embodiments of the invention provide a kind of image partition method, including：
It is gray level image by target image processing, and obtains the candidate subimage region included in the gray level image；
The filtration treatment of setting strategy is carried out to the candidate subimage region included in the gray level image；
The segmentation result to the target image is obtained according to filtration treatment result.
Second aspect, the embodiments of the invention provide a kind of image segmenting device, including：
Image processing module, for being gray level image by target image processing, and obtain what is included in the gray level image
Candidate subimage region；
Candidate's filtering module, for carrying out the mistake of setting strategy to the candidate subimage region included in the gray level image
Filter is handled；
Image segmentation module, for obtaining the segmentation result to the target image according to filtration treatment result.
Technical scheme provided in an embodiment of the present invention, after carrying out gray processing processing to target image, obtain gray-scale map
The candidate subimage region included as in, and the filtration treatment of setting strategy is carried out to candidate subimage region, waited with filtering out
The impurity subgraph in sub-image area is selected, thus reduces the subgraph that mistake is adulterated in target subgraph, improves and cuts figure
Effect.
Brief description of the drawings
Fig. 1 is a kind of flow chart for image partition method that the embodiment of the present invention one provides；
Fig. 2 is a kind of flow chart for image partition method that the embodiment of the present invention two provides；
Fig. 3 is a kind of flow chart for image partition method that the embodiment of the present invention three provides；
Fig. 4 is a kind of structure chart for image segmenting device that the embodiment of the present invention four provides.
Embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention, rather than limitation of the invention.It also should be noted that in order to just
Part related to the present invention rather than entire infrastructure are illustrate only in description, accompanying drawing.
Embodiment one
Fig. 1 is a kind of flow chart for image partition method that the embodiment of the present invention one provides.The method of the present embodiment can be with
Performed by image segmenting device, the device can be realized by way of hardware and/or software, and the method for the present embodiment is general
It is applicable to the situation that user wants to split the target subgraph for obtaining including in target image.Referring to Fig. 1, the present embodiment provides
Image partition method can specifically include it is as follows：
S11, by target image processing it is gray level image, and obtains the candidate subimage area included in the gray level image
Domain.
In the present embodiment, target image can be medical image and engineering image etc..Specifically, by target image
It is gray level image that gray processing processing, which is carried out, by target image processing, and can be by carrying out rim detection and wheel to gray level image
Exterior feature search obtains the candidate subimage region included in gray level image, and wherein candidate subimage region refers to pre- in target image
Phase is the region of target subgraph.
It should be noted that being not especially limited in the present embodiment to edge detection method and Contour searching method, such as may be used
To carry out rim detection to gray level image using Sobel Operator (Sobel operators)；Such as can be by the way that network model be superimposed on
Gray level image, and the energy feature at the grid edge being superimposed in the grid model of gray level image is calculated, according to what is be calculated
Energy feature screening grid edge, and the candidate subimage area included in gray level image is obtained according to the grid edge after screening
Domain.
S12, the filtration treatment that setting strategy is carried out to the candidate subimage region included in the gray level image.
In the present embodiment, setting strategy is used to filter out the impurity sub-image area in candidate subimage region.Specifically,
It can filter out that area is too small, black picture element is excessive by setting strategy or the impurity sub-image regions caused by reason such as border unobvious
Domain.It should be noted that the embodiment of the present invention to setting strategy and being not especially limited only needs that candidate subimage area can be filtered out
Impurity sub-image area in domain, it is as at least one in strategy including but not limited to following filtering policy in set：Area
Filtering policy, stochastic sampling strategy, border extension strategy, similar area consolidation strategy, completely include region merging technique strategy and friendship
Pitch regional processing strategy.
S13, foundation filtration treatment result obtain the segmentation result to the target image.
Specifically, filtration treatment result is matched with target image, determine to whether there is mesh in filtration treatment result
The target subgraph included in logo image, if in the presence of the segmentation result using target subgraph as target image；Otherwise, need
Again target image is split, and the segmentation knot of target image is obtained based on the candidate subimage region after splitting again
Fruit, such as target image is split using level set (level set) algorithm again.
The technical scheme that the present embodiment provides, after carrying out gray processing processing to target image, obtain in gray level image
Comprising candidate subimage region, and to candidate subimage region carry out setting strategy filtration treatment, with filter out candidate son
Impurity subgraph in image-region, thus reduce the subgraph that mistake is adulterated in target subgraph, improve and cut figure effect.
Exemplary, the tactful filtration treatment of setting is carried out to the candidate subimage region included in the gray level image,
It can include：
Area filtering policy, stochastic sampling strategy, side are carried out to the candidate subimage region included in the gray level image
Boundary's expanding policy, similar area consolidation strategy, completely include region merging technique strategy and intersection region processing strategy at least one
The filtration treatment of kind strategy.
Wherein, area filtering policy is used to avoid the area of target subgraph too small, and stochastic sampling strategy is used to avoid mesh
Black picture element is excessive in mark subgraph, and border extension strategy is used for the border unobvious for avoiding target subgraph, and similar area closes
And strategy is used to avoid multiple similar target subgraphs being present, completely includes region merging technique strategy and be used to avoid a target being present
Subgraph is completely contained in inside another target subgraph, and intersection region processing strategy is used to avoid the target subgraph in the presence of intersecting
Picture.
Exemplary, after obtaining the segmentation result to the target image according to filtration treatment result, including：
If not obtaining the target subgraph included in the target image, processing is zoomed in and out to the target image；
The target image after scaling is split using level set algorithm, to obtain candidate subimage region；
The candidate subimage region obtained to level set algorithm carries out the filtration treatment of setting strategy, to obtain to the mesh
The target subgraph included in logo image.
Specifically, if above-mentioned image partition method does not obtain the target subgraph included in target image, using level
Set algorithm is split to target image again, the candidate subimage region included in target image is retrieved, to improve figure
As segmentation precision.The main thought of level set algorithm is that to be considered as level set function be zero the curve of two dimensional image epigraph segmentation
Curve, image splits the process of the solution of equation that a solved function is zero of becoming.For example, Initialize installation one is initial
Curvilinear function (is arranged to gridiron pattern form), and length of curve bound term, area under the curve bound term and curve are added in iterative process
Inside and outside gray scale difference constraint, the process of Optimized model function is just to solve for the mistake of optimal curve plane under multiple constraintss
Journey, finally solve the border that optimal curvilinear function can obtain candidate subimage region equal to curve corresponding to zero.
Also, by reducing the size of target image before image segmentation is carried out using level set algorithm, such as by target
The length and width of image are zoomed within 80 pixels, and the amount of calculation on the one hand reducing level set algorithm is imitated so as to improve image segmentation
Rate, the extrinsic region in candidate subimage region is on the other hand avoided by image scaling, improve the quality of image segmentation.
It should be noted that the candidate subimage region obtained to level set algorithm carries out the filtration treatment of setting strategy
Afterwards, filtration treatment result is matched with target image, if still not obtaining target subgraph, filters out the mistake of cutting failure
Result is filtered, to avoid polluting sub-graph data.
Embodiment two
The present embodiment provides a kind of new image partition method on the basis of above-described embodiment one.In the present embodiment
Specifically provide the implementation of area filtering policy, stochastic sampling strategy and border extension strategy.To correlation in the present embodiment
Threshold value is not especially limited.
Fig. 2 is a kind of flow chart for image partition method that the embodiment of the present invention two provides.Referring to Fig. 2, the present embodiment carries
The image partition method of confession can specifically include as follows：
S21, by target image processing it is gray level image, and obtains the candidate subimage area included in the gray level image
Domain.
, can be with after candidate subimage region is obtained for the ease of subsequently carrying out filtration treatment to candidate subimage region
Take the boundary rectangle frame in each candidate subimage region.If also, need to filter out a certain candidate subimage region in subsequent operation,
The candidate subimage region can be deducted from gray level image, can also be by removing the boundary rectangle in the candidate subimage region
Frame filters out the candidate subimage region to characterize.
If S22, the area in candidate subimage region are less than default area threshold, the candidate subimage area is filtered out
Domain.
In addition, if the area in candidate subimage region is equal to or more than area threshold, retain candidate subimage region.
In the present embodiment, area threshold can be 10000 pixels.By the way that area filtering policy can filter area be especially small does not meet
It is required that candidate subimage region.It should be noted that because area filtering policy efficiency is higher, the operation of area filtering policy can
To be performed before the operation of other filtering policys, to first pass through the quantity in area filtering policy reduction candidate subimage region to subtract
The workload of other few filtering policys.
S23, determine the first extended area for being included in candidate subimage region, wherein first extended area be positioned at
Predetermined width region among candidate subimage region.
Specifically, in the boundary rectangle frame in candidate subimage region, the midpoint of the short side of connection boundary rectangle frame two is formed
Reference line；By reference line towards two long side directions respectively extend one fixed width (such as 10 pixels) formed two extended reference lines, two
Extended reference line and two short sides surround the first extended area of predetermined width (such as 20 pixels).If for example, boundary rectangle frame
It is tall and big to form reference line in width, then two wide midpoints for connecting boundary rectangle frame, reference line is respectively expanded towards two high directions respectively
10 pixels of exhibition form two extended reference lines, and two extended reference lines and two wide surround the first extended area；If boundary rectangle
Frame is wider than height, then two high midpoints for connecting boundary rectangle frame form reference line, by reference line respectively towards two cross directions
Each 10 pixels of extension form two extended reference lines, and two extended reference lines and two height surround the first extended area.
If pixel value is less than the pixel of the first pixel threshold in S24, first extended area, first extension is accounted for
The ratio value in region is more than the first proportion threshold value, then filters out the candidate subimage region belonging to first extended area.
Wherein, the first pixel threshold can be 5, and the first proportion threshold value can be 0.88.By S23 and S24 provide with
The operation of machine sampling policy can filter out the excessive candidate subimage region of black picture element.
S25, the border extension predetermined width by candidate subimage region, to obtain the second of the association of candidate subimage region
Extended area.
Specifically, by the four edges of the boundary rectangle frame in candidate subimage region respectively towards external expansion predetermined width, to obtain
Second extended area of candidate subimage region association.Wherein, predetermined width can be 20 pixels.
If pixel value is more than the pixel of the second pixel threshold in S26, second extended area, second extension is accounted for
The ratio value in region, more than the second proportion threshold value and it is less than the 3rd proportion threshold value, then filters out the second extended area association
Candidate subimage region.
Wherein, second proportion threshold value is less than the 3rd proportion threshold value.Second pixel threshold can be 20, the second ratio
Example threshold value can be 0.35, and the 3rd proportion threshold value can be 0.95.In addition, if the ratio value that S26 is obtained is less than the second ratio threshold
It is worth or more than the 3rd proportion threshold value, then retains the candidate subimage region that the second extended area associates.Carried by S25 and S26
The border extension strategy of confession can filter out candidate subimage region of the border white between excessive and very few, that is, filter out border
Unconspicuous candidate subimage region.
S27, foundation filtration treatment result obtain the segmentation result to the target image.
The technical scheme that the present embodiment provides, after carrying out gray processing processing to target image, obtain in gray level image
Comprising candidate subimage region, and using area filtering policy, stochastic sampling strategy and border extension strategy, to candidate's
Image-region carries out filtration treatment, to filter out the impurity subgraph in candidate subimage region, thus reduces target subgraph
The subgraph of middle doping mistake, improves and cuts figure effect.
Embodiment three
The present embodiment provides a kind of new image partition method on the basis of above-described embodiment one and embodiment two.
Similar area consolidation strategy is specifically provided in the present embodiment, completely include region merging technique strategy and intersection region processing strategy
Implementation.Fig. 3 is a kind of flow chart for image partition method that the embodiment of the present invention three provides.Referring to Fig. 3, the present embodiment carries
The image partition method of confession can specifically include as follows：
S31, by target image processing it is gray level image, and obtains the candidate subimage area included in the gray level image
Domain.
S32, the repetition area for calculating the candidate subimage region intersected, account for the ratio in the candidate subimage region of each intersection
Example value.
Specifically, obtaining the candidate subimage region intersected, and the repetition area in the candidate subimage region of intersection is calculated,
And repeat the ratio value that area accounts for the candidate subimage region of each intersection, i.e. calculate the candidate subimage region of intersection
Area repeats coverage rate, and the similarity in the higher candidate subimage region for representing to intersect of area repetition coverage rate is higher.
If S33, each ratio value being calculated are all higher than the 4th proportion threshold value, retain the candidate subimage area of intersection
A candidate subimage region in domain.
Specifically, so that the 4th proportion threshold value is 0.8 as an example, if in the candidate subimage region of one group of intersection, repeating area
The ratio value for accounting for each candidate subimage region is all higher than 0.8, it is determined that and the candidate subimage region that the group is intersected is similar, and
Retain any one candidate subimage region in the candidate subimage region of group intersection.The similar area provided by S32 and S33
Domain consolidation strategy can avoid retaining multiple areas candidate subimage region similar to position.It should be noted that similar area
Domain consolidation strategy can perform before region merging technique strategy and intersection region processing strategy is completely included, similar by only retaining
One in candidate subimage region reduces the quantity in candidate subimage region, to reduce subsequently completely includes region merging technique plan
Slightly and intersection region handles the amount of calculation of strategy.
S34, acquisition are completely contained in the small candidate subimage region in big candidate subimage region.
If S35, the area of small boundary rectangle frame in the small candidate subimage region account for the big candidate subimage region
Big boundary rectangle frame area ratio value, more than the 5th proportion threshold value, then filter out the small candidate subimage region.
If the ladder on four summits corresponding to four summits to the small boundary rectangle frame of S36, the big boundary rectangle frame
Angle value is all higher than Grads threshold, and the area of the small boundary rectangle frame account for the area of the big boundary rectangle frame ratio value it is big
In the 6th proportion threshold value, then the small candidate subimage region is filtered out；Otherwise, the big candidate subimage region is filtered out.
Wherein, the 5th proportion threshold value is more than the 6th proportion threshold value.5th proportion threshold value can be 0.8, gradient
Threshold value can be 50, and the 6th proportion threshold value can be 0.02.By S34-S36 provide completely include region merging technique strategy can
Avoid having between target subgraph completely include relation caused by impurity subgraph.
S37, the boundary rectangle frame in candidate subimage region to there is cross section carry out descending sort by length-width ratio.
S38, the repetition area value for calculating the high boundary rectangle frame of the length-width ratio boundary rectangle frame low with the length-width ratio intersected.
If S39, the repetition area value being calculated are more than the default value times of the low boundary rectangle frame area of length-width ratio,
Filter the candidate subimage region belonging to the high boundary rectangle frame of length-width ratio.
Wherein, default value can be 0.7 times again.The intersection region processing strategy carried by S37-S39 can avoid
Impurity subgraph caused by intersection candidates sub-image area.
S310, foundation filtration treatment result obtain the segmentation result to the target image.
The technical scheme that the present embodiment provides, after carrying out gray processing processing to target image, obtain in gray level image
Comprising candidate subimage region, and using similar area consolidation strategy, completely include region merging technique strategy and intersection region
Processing strategy, filtration treatment is carried out to candidate subimage region, to filter out the impurity subgraph in candidate subimage region, thus
Reduce the subgraph that mistake is adulterated in target subgraph, improve and cut figure effect.
Example IV
Fig. 4 is a kind of structure chart for image segmenting device that the embodiment of the present invention four provides.The device is generally applicable to
User wants the situation for splitting the target subgraph for obtaining being included in target image.Referring to Fig. 4, the image point that the present embodiment provides
The concrete structure for cutting device is as follows：
Image processing module 41, for being gray level image by target image processing, and obtain and included in the gray level image
Candidate subimage region；
Candidate's filtering module 42, for carrying out setting strategy to the candidate subimage region included in the gray level image
Filtration treatment；
Image segmentation module 43, for obtaining the segmentation result to the target image according to filtration treatment result.
Exemplary, candidate's filtering module 42 specifically can be used for：
Area filtering policy, stochastic sampling strategy, side are carried out to the candidate subimage region included in the gray level image
Boundary's expanding policy, similar area consolidation strategy, completely include region merging technique strategy and intersection region processing strategy at least one
The filtration treatment of kind strategy.
Exemplary, candidate's filtering module 42 specifically can be used for：
If the area in candidate subimage region is less than default area threshold, the candidate subimage region is filtered out.
Exemplary, candidate's filtering module 42 specifically can be used for：
The first extended area included in candidate subimage region is determined, wherein first extended area is positioned at candidate
Predetermined width region among sub-image area；
If pixel value is less than the pixel of the first pixel threshold in first extended area, first extended area is accounted for
Ratio value be more than the first proportion threshold value, then filter out the candidate subimage region belonging to first extended area.
Exemplary, candidate's filtering module 42 specifically can be used for：
By the border extension predetermined width in candidate subimage region, extended with obtaining the second of the association of candidate subimage region
Region；
If pixel value is more than the pixel of the second pixel threshold in second extended area, second extended area is accounted for
Ratio value, more than the second proportion threshold value and be less than the 3rd proportion threshold value, then filter out the candidate of second extended area association
Sub-image area, second proportion threshold value are less than the 3rd proportion threshold value.
Exemplary, candidate's filtering module 42 specifically can be used for：
The repetition area in the candidate subimage region intersected is calculated, accounts for the ratio in the candidate subimage region of each intersection
Value；
If each ratio value being calculated is all higher than the 4th proportion threshold value, in the candidate subimage region for retaining intersection
A candidate subimage region.
Exemplary, candidate's filtering module 42 specifically can be used for：
Obtain the small candidate subimage region being completely contained in big candidate subimage region；
If the area of the small boundary rectangle frame in the small candidate subimage region accounts for the big of the big candidate subimage region
The ratio value of the area of boundary rectangle frame, more than the 5th proportion threshold value, then filter out the small candidate subimage region；
If the Grad on four summits corresponding to four summits of the big boundary rectangle frame to the small boundary rectangle frame
It is all higher than Grads threshold, and the area of the small boundary rectangle frame accounts for the ratio value of the area of the big boundary rectangle frame and is more than the
Six proportion threshold values, then filter out the small candidate subimage region；Otherwise, the big candidate subimage region is filtered out, wherein described
5th proportion threshold value is more than the 6th proportion threshold value.
Exemplary, candidate's filtering module 42 specifically can be used for：
To there is the boundary rectangle frame in the candidate subimage region of cross section to carry out descending sort by length-width ratio；
Calculate the repetition area value of the high boundary rectangle frame of the length-width ratio boundary rectangle frame low with the length-width ratio intersected；
If the repetition area value being calculated is more than the default value times of the low boundary rectangle frame area of length-width ratio, filter
Candidate subimage region belonging to the high boundary rectangle frame of length-width ratio.
Exemplary, above-mentioned image segmenting device can include：
Image scaling module, after obtaining to the segmentation result of the target image in foundation filtration treatment result,
If not obtaining the target subgraph included in the target image, processing is zoomed in and out to the target image；
Level-set segmentation module, for being split using level set algorithm to the target image after scaling, to be waited
Select sub-image area；
Target subgraph module, the candidate subimage region for being obtained to level set algorithm carry out the filtering of setting strategy
Processing, to obtain the target subgraph to being included in the target image.
The image segmenting device that the present embodiment provides, the image partition method provided with any embodiment of the present invention belong to
Same inventive concept, the image partition method that any embodiment of the present invention is provided is can perform, possesses execution image partition method
Corresponding functional module and beneficial effect.Not ins and outs of detailed description in the present embodiment, reference can be made to the present invention is any real
The image partition method of example offer is provided.
Pay attention to, above are only presently preferred embodiments of the present invention and institute's application technology principle.It will be appreciated by those skilled in the art that
The invention is not restricted to specific embodiment described here, can carry out for a person skilled in the art various obvious changes,
Readjust and substitute without departing from protection scope of the present invention.Therefore, although being carried out by above example to the present invention
It is described in further detail, but the present invention is not limited only to above example, without departing from the inventive concept, also
Other more equivalent embodiments can be included, and the scope of the present invention is determined by scope of the appended claims.
Claims (18)
1. a kind of image partition method, including：
It is gray level image by target image processing, and obtains the candidate subimage region included in the gray level image；
The filtration treatment of setting strategy is carried out to the candidate subimage region included in the gray level image；
The segmentation result to the target image is obtained according to filtration treatment result.
2. according to the method for claim 1, wherein, the candidate subimage region included in the gray level image is set
The filtration treatment of fixed strategy, including：
Area filtering policy, stochastic sampling strategy, border expansion are carried out to the candidate subimage region included in the gray level image
Exhibition strategy, similar area consolidation strategy, completely include at least one of region merging technique strategy and intersection region processing strategy plan
Filtration treatment slightly.
3. according to the method for claim 2, wherein, face is carried out to the candidate subimage region included in the gray level image
The filtration treatment of product filtering policy, including：
If the area in candidate subimage region is less than default area threshold, the candidate subimage region is filtered out.
4. according to the method for claim 2, wherein, the candidate subimage region that is included in the gray level image is carried out with
The filtration treatment of machine sampling policy, including：
The first extended area included in candidate subimage region is determined, wherein first extended area is positioned at candidate's subgraph
As the predetermined width region among region；
If pixel value is less than the pixel of the first pixel threshold in first extended area, the ratio of first extended area is accounted for
Example value is more than the first proportion threshold value, then filters out the candidate subimage region belonging to first extended area.
5. according to the method for claim 2, wherein, side is carried out to the candidate subimage region included in the gray level image
The filtration treatment of boundary's expanding policy, including：
By the border extension predetermined width in candidate subimage region, to obtain the second expansion area of candidate subimage region association
Domain；
If pixel value is more than the pixel of the second pixel threshold in second extended area, the ratio of second extended area is accounted for
Example value, more than the second proportion threshold value and is less than the 3rd proportion threshold value, then filters out candidate's subgraph of the second extended area association
As region, second proportion threshold value is less than the 3rd proportion threshold value.
6. according to the method for claim 2, wherein, phase is carried out to the candidate subimage region included in the gray level image
Like the filtration treatment of region merging technique strategy, including：
The repetition area in the candidate subimage region intersected is calculated, accounts for the ratio value in the candidate subimage region of each intersection；
If each ratio value being calculated is all higher than the 4th proportion threshold value, retain one in the candidate subimage region of intersection
Individual candidate subimage region.
7. according to the method for claim 2, wherein, the candidate subimage region included in the gray level image is carried out
The filtration treatment of full inclusion region consolidation strategy, including：
Obtain the small candidate subimage region being completely contained in big candidate subimage region；
If the area of the small boundary rectangle frame in the small candidate subimage region accounts for the big external of the big candidate subimage region
The ratio value of the area of rectangle frame, more than the 5th proportion threshold value, then filter out the small candidate subimage region；
If the Grad on four summits is big corresponding to four summits of the big boundary rectangle frame to the small boundary rectangle frame
The ratio value that the area of the big boundary rectangle frame is accounted in the area of Grads threshold, and the small boundary rectangle frame is more than the 6th ratio
Example threshold value, then filter out the small candidate subimage region；Otherwise, the big candidate subimage region is filtered out, wherein the described 5th
Proportion threshold value is more than the 6th proportion threshold value.
8. according to the method for claim 2, wherein, the candidate subimage region included in the gray level image is handed over
The filtration treatment of regional processing strategy is pitched, including：
To there is the boundary rectangle frame in the candidate subimage region of cross section to carry out descending sort by length-width ratio；
Calculate the repetition area value of the high boundary rectangle frame of the length-width ratio boundary rectangle frame low with the length-width ratio intersected；
If the repetition area value being calculated is more than the default value times of the low boundary rectangle frame area of length-width ratio, length and width are filtered
Than the candidate subimage region belonging to high boundary rectangle frame.
9. according to the method described in claim any one of 1-8, wherein, obtained according to filtration treatment result to the target image
Segmentation result after, in addition to：
If not obtaining the target subgraph included in the target image, processing is zoomed in and out to the target image；
The target image after scaling is split using level set algorithm, to obtain candidate subimage region；
The candidate subimage region obtained to level set algorithm carries out the filtration treatment of setting strategy, to obtain to the target figure
The target subgraph included as in.
10. a kind of image segmenting device, including：
Image processing module, for being gray level image by target image processing, and obtain the candidate included in the gray level image
Sub-image area；
Candidate's filtering module, for being carried out to the candidate subimage region included in the gray level image at the tactful filtering of setting
Reason；
Image segmentation module, for obtaining the segmentation result to the target image according to filtration treatment result.
11. device according to claim 10, wherein, candidate's filtering module is specifically used for：
Area filtering policy, stochastic sampling strategy, border expansion are carried out to the candidate subimage region included in the gray level image
Exhibition strategy, similar area consolidation strategy, completely include at least one of region merging technique strategy and intersection region processing strategy plan
Filtration treatment slightly.
12. device according to claim 11, wherein, candidate's filtering module is specifically used for：
If the area in candidate subimage region is less than default area threshold, the candidate subimage region is filtered out.
13. device according to claim 11, wherein, candidate's filtering module is specifically used for：
The first extended area included in candidate subimage region is determined, wherein first extended area is positioned at candidate's subgraph
As the predetermined width region among region；
If pixel value is less than the pixel of the first pixel threshold in first extended area, the ratio of first extended area is accounted for
Example value is more than the first proportion threshold value, then filters out the candidate subimage region belonging to first extended area.
14. device according to claim 11, wherein, candidate's filtering module is specifically used for：
By the border extension predetermined width in candidate subimage region, to obtain the second expansion area of candidate subimage region association
Domain；
If pixel value is more than the pixel of the second pixel threshold in second extended area, the ratio of second extended area is accounted for
Example value, more than the second proportion threshold value and is less than the 3rd proportion threshold value, then filters out candidate's subgraph of the second extended area association
As region, second proportion threshold value is less than the 3rd proportion threshold value.
15. device according to claim 11, wherein, candidate's filtering module is specifically used for：
The repetition area in the candidate subimage region intersected is calculated, accounts for the ratio value in the candidate subimage region of each intersection；
If each ratio value being calculated is all higher than the 4th proportion threshold value, retain one in the candidate subimage region of intersection
Individual candidate subimage region.
16. device according to claim 11, wherein, candidate's filtering module is specifically used for：
Obtain the small candidate subimage region being completely contained in big candidate subimage region；
If the area of the small boundary rectangle frame in the small candidate subimage region accounts for the big external of the big candidate subimage region
The ratio value of the area of rectangle frame, more than the 5th proportion threshold value, then filter out the small candidate subimage region；
If the Grad on four summits is big corresponding to four summits of the big boundary rectangle frame to the small boundary rectangle frame
The ratio value that the area of the big boundary rectangle frame is accounted in the area of Grads threshold, and the small boundary rectangle frame is more than the 6th ratio
Example threshold value, then filter out the small candidate subimage region；Otherwise, the big candidate subimage region is filtered out, wherein the described 5th
Proportion threshold value is more than the 6th proportion threshold value.
17. device according to claim 11, wherein, candidate's filtering module is specifically used for：
To there is the boundary rectangle frame in the candidate subimage region of cross section to carry out descending sort by length-width ratio；
Calculate the repetition area value of the high boundary rectangle frame of the length-width ratio boundary rectangle frame low with the length-width ratio intersected；
If the repetition area value being calculated is more than the default value times of the low boundary rectangle frame area of length-width ratio, length and width are filtered
Than the candidate subimage region belonging to high boundary rectangle frame.
18. according to the device described in claim any one of 10-17, in addition to：
Image scaling module, after obtaining to the segmentation result of the target image in foundation filtration treatment result, if not
The target subgraph included in the target image is obtained, then processing is zoomed in and out to the target image；
Level-set segmentation module, for being split using level set algorithm to the target image after scaling, to obtain candidate's
Image-region；
Target subgraph module, the candidate subimage region for being obtained to level set algorithm are carried out at the filtering of setting strategy
Reason, to obtain the target subgraph to being included in the target image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610377503.XA CN107452002A (en) | 2016-05-31 | 2016-05-31 | A kind of image partition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610377503.XA CN107452002A (en) | 2016-05-31 | 2016-05-31 | A kind of image partition method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107452002A true CN107452002A (en) | 2017-12-08 |
Family
ID=60485062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610377503.XA Pending CN107452002A (en) | 2016-05-31 | 2016-05-31 | A kind of image partition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107452002A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171716A (en) * | 2017-12-25 | 2018-06-15 | 北京奇虎科技有限公司 | Video personage based on the segmentation of adaptive tracing frame dresss up method and device |
| CN109325943A (en) * | 2018-09-10 | 2019-02-12 | 深圳开立生物医疗科技股份有限公司 | A kind of three-D volumes measurement method and device |
| CN109816011A (en) * | 2019-01-21 | 2019-05-28 | 厦门美图之家科技有限公司 | Generate the method and video key frame extracting method of portrait parted pattern |
| CN109978810A (en) * | 2017-12-26 | 2019-07-05 | 柴岗 | Detection method, system, equipment and the storage medium of mole |
| CN110782463A (en) * | 2019-10-31 | 2020-02-11 | 京东方科技集团股份有限公司 | Method and device for determining division mode, display method and equipment and storage medium |
| CN110827245A (en) * | 2019-10-28 | 2020-02-21 | 上海悦易网络信息技术有限公司 | Method and equipment for detecting screen display disconnection |
| CN112070788A (en) * | 2020-09-03 | 2020-12-11 | 武汉工程大学 | Image processing method for fast statistical deformation twinning based on block gradient segmentation |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663354A (en) * | 2012-03-26 | 2012-09-12 | 腾讯科技（深圳）有限公司 | Face calibration method and system thereof |
| CN103235938A (en) * | 2013-05-03 | 2013-08-07 | 北京国铁华晨通信信息技术有限公司 | Method and system for detecting and identifying license plate |
| CN104851085A (en) * | 2014-02-17 | 2015-08-19 | 征图新视（江苏）科技有限公司 | Method and system automatically obtaining detection zone in image |
| JP2015228113A (en) * | 2014-05-30 | 2015-12-17 | キヤノン株式会社 | Image processor and image processing method |
| CN105488468A (en) * | 2015-11-26 | 2016-04-13 | 浙江宇视科技有限公司 | Method and device for positioning target area |
-
2016
- 2016-05-31 CN CN201610377503.XA patent/CN107452002A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663354A (en) * | 2012-03-26 | 2012-09-12 | 腾讯科技（深圳）有限公司 | Face calibration method and system thereof |
| CN103235938A (en) * | 2013-05-03 | 2013-08-07 | 北京国铁华晨通信信息技术有限公司 | Method and system for detecting and identifying license plate |
| CN104851085A (en) * | 2014-02-17 | 2015-08-19 | 征图新视（江苏）科技有限公司 | Method and system automatically obtaining detection zone in image |
| JP2015228113A (en) * | 2014-05-30 | 2015-12-17 | キヤノン株式会社 | Image processor and image processing method |
| CN105488468A (en) * | 2015-11-26 | 2016-04-13 | 浙江宇视科技有限公司 | Method and device for positioning target area |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171716A (en) * | 2017-12-25 | 2018-06-15 | 北京奇虎科技有限公司 | Video personage based on the segmentation of adaptive tracing frame dresss up method and device |
| CN109978810A (en) * | 2017-12-26 | 2019-07-05 | 柴岗 | Detection method, system, equipment and the storage medium of mole |
| CN109978810B (en) * | 2017-12-26 | 2024-03-12 | 南通罗伯特医疗科技有限公司 | Method, system, device and storage medium for detecting nevus |
| CN109325943A (en) * | 2018-09-10 | 2019-02-12 | 深圳开立生物医疗科技股份有限公司 | A kind of three-D volumes measurement method and device |
| CN109816011A (en) * | 2019-01-21 | 2019-05-28 | 厦门美图之家科技有限公司 | Generate the method and video key frame extracting method of portrait parted pattern |
| CN109816011B (en) * | 2019-01-21 | 2021-09-07 | 厦门美图之家科技有限公司 | Video key frame extraction method |
| CN110827245A (en) * | 2019-10-28 | 2020-02-21 | 上海悦易网络信息技术有限公司 | Method and equipment for detecting screen display disconnection |
| CN110782463A (en) * | 2019-10-31 | 2020-02-11 | 京东方科技集团股份有限公司 | Method and device for determining division mode, display method and equipment and storage medium |
| CN110782463B (en) * | 2019-10-31 | 2022-06-24 | 京东方科技集团股份有限公司 | Method and device for determining division mode, display method and equipment and storage medium |
| CN112070788A (en) * | 2020-09-03 | 2020-12-11 | 武汉工程大学 | Image processing method for fast statistical deformation twinning based on block gradient segmentation |
| CN112070788B (en) * | 2020-09-03 | 2023-03-10 | 武汉工程大学 | Image Processing Method for Fast Statistical Deformation Twins Based on Block Gradient Segmentation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107452002A (en) | A kind of image partition method and device | |
| CN105260713B (en) | A kind of method for detecting lane lines and device | |
| JP4172941B2 (en) | Land parcel data creation method and apparatus | |
| CN102999886A (en) | Image edge detector and ruler raster grid line precision detection system | |
| CN105987684A (en) | Monocular vision-based agricultural vehicle navigation line detection system and method | |
| CN107610092B (en) | Pavement crack dynamic detection method based on video stream | |
| CN101681432A (en) | Photo-document segmentation method and system | |
| CN110135311A (en) | A Pore and Pore Throat Identification System and Method Based on 3D Core Scanning Image | |
| CN109214439A (en) | A kind of infrared image icing River detection method based on multi-feature fusion | |
| CN108961230B (en) | Identification and extraction method for structural surface crack characteristics | |
| CN107729814A (en) | A kind of method and device for detecting lane line | |
| CN106971185A (en) | A kind of license plate locating method and device based on full convolutional network | |
| CN106504261B (en) | A kind of image partition method and device | |
| CN111507986B (en) | Method and device for determining internal wave and condition of pore | |
| CN106934829A (en) | The detection method and system of a kind of surface crack | |
| CN112085699A (en) | Pavement crack extraction method based on two-dimensional image | |
| CN103208121B (en) | Based on the remote sensing image segmentation method that bounds constraint merges with two benches | |
| CN104778710A (en) | A method of edge detection in morphological images based on quantum theory | |
| CN108537787A (en) | A kind of quality judging method of facial image | |
| CN103606179B (en) | Animation image display method and device | |
| CN108154471A (en) | A kind of method and apparatus for remote sensing image splicing | |
| CN113989673A (en) | Coastline automatic extraction method based on domestic high-score data and deep learning model | |
| CN109766892A (en) | A method for character segmentation of ray detection image marking information based on edge detection | |
| Champion et al. | A test of automatic building change detection approaches | |
| CN102760295A (en) | Fire disaster image detection system for edge detection-based operator |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171208 |