CN108305268A - A kind of image partition method and device - Google Patents
A kind of image partition method and device Download PDFInfo
- Publication number
- CN108305268A CN108305268A CN201810005450.8A CN201810005450A CN108305268A CN 108305268 A CN108305268 A CN 108305268A CN 201810005450 A CN201810005450 A CN 201810005450A CN 108305268 A CN108305268 A CN 108305268A
- Authority
- CN
- China
- Prior art keywords
- image
- split
- pixel
- segmentation
- entropy
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
This application discloses a kind of image partition method and device, this method includes：Image to be split is first obtained, the image to be split includes object to be split；Image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture is the corresponding comentropy of the pixel；The unclosed region of object to be split in entropy diagram picture is closed, obtains dividing preceding image；The object to be split is partitioned into from image before segmentation.It can be seen that, image to be split is converted into entropy diagram picture by using comentropy, then profile Seal treatment is carried out to the object to be split in image again, this pretreatment mode, the image outline that cutting object can be treated carries out enhancing processing, accurately highlights the marginal position of object to be split, realizes image segmentation on this basis, it more can be accurately partitioned into object to be split, improve the accuracy of image segmentation result.
Description
Technical field
This application involves technical field of image processing more particularly to a kind of image partition methods and device.
Background technology
The purpose of medical image segmentation be by extract medical image feature, object of interest (such as certain organ) from
It is separated in medical image, the purpose is to analyze and calculate the dissection of object of interest, pathology, physiology, physics etc.
Information, correct segmentation result can help doctor to provide correct diagnostic result.Due to the practical acquisition condition of medical image
It is often different, cause medical image extremely complex, this makes medical image segmentation have certain complexity.And medicine figure
As the selection of dividing method, it is largely dependent upon the particularity, image imaging mode and influence image of image itself
The human factor (for example human body moves) of image quality and factor (such as noise of equipment) can not be resisted etc., these all can be very
Subsequent image segmentation is influenced in big degree.
Currently, widely used medical image cutting method both at home and abroad, includes dividing method based on region, is based on edge
Dividing method, in conjunction with the dividing method (such as artificial neural network method, deformable model method) of Specific Theory Tools, based on small
The dividing method of wave conversion, based on statistical dividing method, the dividing method based on mathematical morphology, based on region growing
Dividing method, the dividing method based on Markov random field, the dividing method based on fuzzy theory and be based on active contour
Dividing method of model, etc..
However, each above-mentioned dividing method be all zero noise without degenerate, the profile closure of object to be split is good and
In the apparent image basis of grey scale change, ideal segmentation effect can be obtained, but when medical image itself it is second-rate when,
These methods will show bad as a result, the result of segmentation is inaccurate or can not simply divide.That is, doctor
Learn image segmentation before quality be only determine segmentation effect quality key factor, therefore, it is necessary to before image segmentation to original
Medical image is pre-processed, and the factor for influencing segmentation effect is eliminated as much as, and is the committed step for obtaining ideal segmentation effect.
For this purpose, there is researcher's proposition, original image is subjected to binary conversion treatment, and combine mathematics binary morphology to original
The fringe region of object of interest optimizes processing in beginning image, at this point, on image each pixel gray value be 0 or
255, image segmentation is carried out based on this grey value difference.But the image after binaryzation can not remain hidden between pixel
The information of minor variations, this position of edge of object of interest in the picture is displaced (i.e. edge positioning is inaccurate
Really), so as to cause segmentation result inaccuracy.
Invention content
The main purpose of the embodiment of the present application is to provide a kind of image partition method and device, can improve image segmentation
As a result accuracy.
The embodiment of the present application provides a kind of image partition method, including：
Image to be split is obtained, the image to be split includes object to be split；
The image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture is the pixel
The corresponding comentropy of point；
The unclosed region of the object to be split in the entropy diagram picture is closed, obtains dividing preceding image；
The object to be split is partitioned into from image before the segmentation.
Optionally, described that the image to be split is converted into entropy diagram picture, including：
Using each pixel in the image to be split as target pixel points, the patch of default size is utilized
Block selects the target pixel points in the image center to be split；
According to the gray value of each pixel in the patch block, the corresponding comentropy of the target pixel points is calculated；
Judge whether that the corresponding comentropy of each pixel is equal at least one pixel region and the pixel region
It is zero；
If so, increase the size of the patch block, continues to execute and described select the mesh in the image center to be split
The step of marking pixel；
If it is not, then utilizing the corresponding comentropy formation entropy image of each pixel in the image to be split.
Optionally, the unclosed region by the object to be split in the entropy diagram picture is closed, including：
Opening operation using mathematical morphology or closed operation, by the unclosed of the object to be split in the entropy diagram picture
It is closed in region.
Optionally, described to be partitioned into the object to be split from image before the segmentation, including：
Using the dividing method based on movable contour model, it is described to be split right to be partitioned into from image before the segmentation
As.
Optionally, after the acquisition image to be split, including：
Denoising is carried out to the image to be split by the way of shearing wave denoising, obtains the image to be split after denoising.
Optionally, after the image to be split obtained after denoising, further include：
Using shock filter, edge enhancing is carried out to the image to be split after the denoising.
Optionally, described to obtain after dividing preceding image, further include：
Gray scale stretching is carried out to image before the segmentation.
The embodiment of the present application also provides a kind of image segmentation devices, including：
Image acquisition unit, for obtaining image to be split, the image to be split includes object to be split；
Image conversion unit, for the image to be split to be converted to entropy diagram picture, each pixel of the entropy diagram picture
Pixel value be the corresponding comentropy of the pixel；
Border seal unit, for the unclosed region of the object to be split in the entropy diagram picture to be closed,
It obtains dividing preceding image；
Image segmentation unit, for being partitioned into the object to be split from image before the segmentation.
Optionally, described image converting unit includes：
Patch frame selects subelement, for using each pixel in the image to be split as target pixel points,
Using the patch block of default size, the target pixel points are selected in the image center to be split；
Entropy computation subunit calculates the target picture for the gray value according to each pixel in the patch block
The corresponding comentropy of vegetarian refreshments；
Entropy judgment sub-unit, for judging whether each picture at least one pixel region and the pixel region
The corresponding comentropy of vegetarian refreshments is zero；
Entropy is reruned subelement, if for there are each pixels pair at least one pixel region and the pixel region
The comentropy answered is zero, then increases the size of the patch block, triggers the patch frame and selects subelement in the figure to be split
As center selects the target pixel points；
Image conversion subunit, if for there is no each pixels at least one pixel region and the pixel region
Corresponding comentropy is zero, then utilizes the corresponding comentropy formation entropy image of each pixel in the image to be split.
The embodiment of the present application also provides a kind of image segmentation devices, including：Processor, memory, system bus；
The processor and the memory are connected by the system bus；
The memory includes instruction, described instruction for storing one or more programs, one or more of programs
The processor is set to execute method described in any one of the above embodiments when being executed by the processor.
A kind of image partition method and device provided by the embodiments of the present application, first obtain image to be split, described to be split
Image includes object to be split；Image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture
For the corresponding comentropy of the pixel；The unclosed region of object to be split in entropy diagram picture is closed, is divided
Preceding image；The object to be split is partitioned into from image before segmentation.As it can be seen that image to be split is converted by using comentropy
For entropy diagram picture, profile Seal treatment then carried out to the object to be split in image again, this pretreatment mode can be treated point
The image outline for cutting object carries out enhancing processing, accurately highlights the marginal position of object to be split, realizes image on this basis
Segmentation, more can accurately be partitioned into object to be split, improve the accuracy of image segmentation result.
Description of the drawings
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the application
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1 is a kind of flow diagram for image partition method that the embodiment of the present application one provides；
Fig. 2 is the unclosed image schematic diagram that the embodiment of the present application one provides；
Fig. 3 is a kind of flow diagram for image partition method that the embodiment of the present application two provides；
Fig. 4 is the entropy image conversion process schematic diagram that the embodiment of the present application two provides；
Fig. 5 is the Morphological scale-space image schematic diagram that the embodiment of the present application two provides；
Fig. 6 A are one of the cutting procedure schematic diagram that the embodiment of the present application two provides；
Fig. 6 B are the two of the cutting procedure schematic diagram that the embodiment of the present application two provides；
Fig. 7 A are the search pattern schematic diagram from the outside to the core that the embodiment of the present application two provides；
Fig. 7 B are the search pattern schematic diagram from inside to outside that the embodiment of the present application two provides；
Fig. 7 C are the two-way search pattern schematic diagram that the embodiment of the present application two provides；
Fig. 8 is a kind of flow diagram for image partition method that the embodiment of the present application three provides；
Fig. 9 is a kind of composition schematic diagram for image segmentation device that the embodiment of the present application four provides；
Figure 10 is a kind of hardware architecture diagram for image segmentation device that the embodiment of the present application five provides.
Specific implementation mode
To keep the purpose, technical scheme and advantage of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, technical solutions in the embodiments of the present application is clearly and completely described, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
The every other embodiment obtained without making creative work, shall fall in the protection scope of this application.
Currently, the extensive use of advanced imaging technique medically greatly improves the quality of medical services, and scheme
The effect to become more and more important is played as being segmented in medical imaging field, medical image segmentation is in clinical diagnosis, pathological analysis
And the application of human body organ three-dimensional reproduction etc. is all vital.Since clinical application is to the accurate of medical image segmentation
Degree require it is larger, although there are many partitioning algorithm, the accuracy of segmentation result is high not enough.Specifically, existing
Each dividing method is all and grey scale change apparent image good in profile closure of zero noise without degeneration, object to be split
On the basis of, can obtain ideal segmentation effect, but when medical image itself it is second-rate when, these methods will be shown
It is bad as a result, the result of segmentation is inaccurate or can not simply divide.That is, quality before the segmentation of medical image
It is only and determines that the key factor of segmentation effect quality to the greatest extent may be used therefore, it is necessary to be pre-processed to original image before image segmentation
Can eliminate the effects of the act the factor of segmentation effect, be the committed step for obtaining ideal segmentation effect.
For this purpose, the embodiment of the present application provides a kind of image partition method, original image to be split is converted into entropy diagram
Picture carries out closure reparation to the profile of object to be split in entropy diagram picture, obtains dividing preceding image, so as to complete dividing later
Pretreatment work before cutting.
In this pretreatment mode, first, entropy diagram picture can remain hidden the information of the minor variations between pixel, you can
To retain the marginal information of object to be split in image, image edge location will not be made to move as existing bianry image
Position, moreover, compared with original image, entropy diagram picture can also highlight the marginal position of object to be split, and can be to picture noise
Play certain inhibiting effect；Secondly, border seal processing then to the object to be split in entropy diagram picture is carried out, it can be further bright
True split position.Based on this pre-processed results, when carrying out image segmentation, can more accurately locating segmentation position, to
It is accurate to realize image segmentation.
It is specifically introduced with reference to Fig. 1-8 pairs of image partition methods provided by the embodiments of the present application.
Embodiment one
Referring to Fig. 1, for a kind of flow diagram for image partition method that the present embodiment one provides.The image segmentation side
Method includes the following steps：
S101：Image to be split is obtained, the image to be split includes object to be split.
It should be noted that for a untreated original image, wish if the original image includes user
The object to be split being partitioned into can integrally regard the original image as the image to be split, can also be by the original image
In subregion (for example the region can be manually intercepted using drawing tools) as the image to be split, as long as described wait for
Divide in image includes the object to be split.
It should also be noted that, the image to be split can be medical image, can also be other types of image, this
Embodiment is not construed as limiting this.For example, when the image to be split is medical image, it is to be split in the image to be split
Object can be human organ, such as heart.
S102：The image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture is institute
State the corresponding comentropy of pixel.
It should be noted that the image to be split is actually an intensity image, the ash of each pixel is described
Intensity value is spent, the edge feature of object to be split in image can not be highlighted.And the present embodiment converts the image to be split
For entropy diagram picture, entropy diagram picture is actually a structural images, the not instead of gray-scale intensity value of description, the structure group in image
At therefore, entropy diagram picture can remain hidden the information of the minor variations between pixel, can be good at the figure for showing object to be split
As edge, and image segmentation is focused on be exactly object to be split profile information, therefore, this is carried out for follow-up image of accurately dividing
Prepare.Further, entropy diagram picture is not bianry image, it also maintains the intensity variation of the image to be split.
It should also be noted that, entropy diagram picture can also inhibit the interference of picture noise to a certain extent.
S103：The unclosed region of the object to be split in the entropy diagram picture is closed, obtains dividing preceding figure
Picture.
In practical application, when object to be split is non-closed in the picture, when especially unclosed region is larger, into
When row image segmentation, it may appear that the problem of object to be split is not partitioned into all.Therefore, the present embodiment can be to be split
The unclosed region of object carries out Seal treatment, to obtain dividing preceding image.
For example, with reference to unclosed image schematic diagram shown in Fig. 2, it is assumed that Fig. 2 left-side images are to be split in entropy diagram picture
Object, the image are unclosed, and the closed object to be split on the right side of Fig. 2 is obtained after being carried out Seal treatment.
S104：The object to be split is partitioned into from image before the segmentation.
The example for continuing above-mentioned Fig. 2 searches the object to be split using partitioning algorithm from image before the segmentation
Closed edge carries out image segmentation on the basis of the closed edge later, is therefrom partitioned into the object to be split.It needs
Bright, existing any type image segmentation algorithm may be used in the present embodiment, and the present embodiment is without limitation.
To sum up, a kind of image partition method provided in this embodiment, first obtains image to be split, in the image to be split
Including object to be split；Image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture is described
The corresponding comentropy of pixel；The unclosed region of object to be split in entropy diagram picture is closed, obtains dividing preceding image；
The object to be split is partitioned into from image before segmentation.As it can be seen that image to be split is converted to entropy diagram by using comentropy
Picture, then carries out profile Seal treatment to the object to be split in image again, and this pretreatment mode can treat cutting object
Image outline carry out enhancing processing, accurately highlight the marginal position of object to be split, realize image segmentation, energy on this basis
It is enough to be more accurately partitioned into object to be split, improve the accuracy of image segmentation result.
Embodiment two
Referring to Fig. 3, for a kind of flow diagram for image partition method that the present embodiment two provides.The image segmentation side
Method includes the following steps：
S301：Image to be split is obtained, the image to be split includes object to be split.
It should be noted that this step S301 is identical as the S101 in above-described embodiment one, related introduction refers to implementation
Example one, details are not described herein.
Following step S302-S306 is the specific implementation of S102 in above-described embodiment one.
S302：Using each pixel in the image to be split as target pixel points, default size is utilized
Patch block selects the target pixel points in the image center to be split.
For each pixel in the image to be split, using the patch block of a default size, (for example square is mended
Fourth block), each pixel is selected in the image center to be split respectively, and keep each pixel homogeneous in patch position in the block
With (for example being respectively positioned on the center of patch block), for ease of being distinguished with other pixels, the present embodiment is by the center
Pixel is known as target pixel points.
Entropy image conversion process schematic diagram shown in Figure 4, now with two different pixels in the image to be split of left side
For point, the two pixels are located at the center of patch block.
S303：According to the gray value of each pixel in the patch block, the corresponding information of the target pixel points is calculated
Entropy.
Patch block where the target pixel points is calculated based on the gray value of each pixel in the patch block
Corresponding grey level histogram (such as histogram shown in Fig. 4).It should be noted that grey level histogram is about grey level distribution
Function, be the statistics to grey level distribution in image, it indicate image in certain gray level pixel number, reflection
The frequency that certain gray scale occurs in image.
Then, according to the gray probability statistical result of grey level histogram, the corresponding comentropy of the target pixel points is calculated,
Comentropy namely Shannon entropy, Shannon entropy calculation formula are as follows：
Wherein, i=1,2 ... N, N are the pixel total number in image to be split；Y=i represents ith pixel point work
For the target pixel points when corresponding patch block region, p represents the density function of variable Y.
S304：Judge whether the corresponding information of each pixel at least one pixel region and the pixel region
Entropy is zero, if so, S305 is executed, if it is not, then executing S306.
S305：The size for increasing the patch block, using the patch block after increase, continue to execute in step S302 "
The image center to be split selects the target pixel points " the step of.
S306：Utilize the corresponding comentropy formation entropy image of each pixel in the image to be split.
About step S304-S306, it should be noted that when patch block is excessive, may cause final entropy diagram as in
The obscurity boundary of object to be split；When patch block is too small, it is zero that may lead to the corresponding comentropy of continuous multiple pixels, from
And occur being not belonging to the boundary line of object to be split, it, can be to the subsequent image based on image boundary point if there is such case
Generation interference is cut, therefore, the patch block of suitable size should be selected.
In the present embodiment, to avoid the obscurity boundary of object to be split, the initial size of patch block will not select it is excessive, but
To avoid patch block is too small from above-mentioned interference occur, relatively small initial patch block can be selected, further according to the entropy of calculating
Constantly current patch block is adjusted.
Therefore, it after the comentropy of each pixel is calculated using above-mentioned S303, is judged whether by S304
One or more special pixel regions, wherein the corresponding comentropy of pixel in each special pixel region is zero；If
It is then to illustrate that current patch block is less than normal, increases the size of current patch block by S305, recalculate the information of each pixel
Entropy；Otherwise, illustrate that current patch block is suitable, can then pass through S306 formation entropy images, such as entropy diagram picture shown in Fig. 4.
S307：Opening operation using mathematical morphology or closed operation, by the object to be split in the entropy diagram picture
Unclosed region is closed, and obtains dividing preceding image.
In the present embodiment, the opening and closing operation that mathematical morphology may be used handles entropy diagram picture, by entropy diagram as in
The profile of object to be split carries out Seal treatment, that is, is attached place unclosed in profile, forms closed contour.
Mathematical morphology refers to the science of shape and structure, and in image processing field, mathematical morphology is analysis chart
The method of inherent geometry as in.Specifically, it first selectes the structural element with definite shape and size and passes through shape later
State student movement calculation can remain the geometric properties similar with structural element of shape and size in entropy diagram picture, and remaining spy
Sign filters out.
By using the opening and closing operation of mathematical morphology, the image outline of the object to be split in entropy diagram picture can be carried out
Repairing, meanwhile, it is capable to which other organization edges in picture noise and reduction entropy diagram picture is inhibited to treat cutting object generation interference.
For ease of understanding, it is now assumed that A is the entropy diagram picture, B is previously selected structural element, below to mathematical morphology
Opening and closing operation be introduced.
Morphology opening operation refers to being expanded again with B after A is corroded by B, that is, A can be denoted as by the morphology opening operation of B
The effect of morphology opening operation is：The subject area of structural element B cannot be included by deleting completely in A, smooth
The profile of object is disconnected narrow connection, eliminates tiny protrusion.For example, at morphology as shown in Figure 5
Image schematic diagram is managed, figure (a) obtains image (b) after morphology opening operation.
Closing operation of mathematical morphology refers to being corroded again with B after A is expanded by B, that is, A can be denoted as by the closing operation of mathematical morphology of B
A·B：
Unlike morphology opening operation, notch narrow in image can generally be connected shape by closing operation of mathematical morphology
At elongated curved mouth, and the hole that packing ratio structural element is small.For example, still as shown in figure 5, figure (a) is through closing operation of mathematical morphology
After obtain image (c), figure (b) obtains image (d) after closing operation of mathematical morphology.
S308：Using the dividing method based on movable contour model, described wait for point is partitioned into from image before the segmentation
Cut object.
Active contour is defined as geometrically being described as one two by the dividing method based on movable contour model
A curve on the dimension space plane of delineation, the every bit on curve are known as snake point, and all snake point straight lines or curve are connected
Get up and constitutes entire profile.In this way, entire profile can be indicated only with these snake points, and the deformation of active contour
It can be completed by the movement of snake point, deformation process is exactly active contour in external energy (external force) and internal energy
Under the action of (internal force), the process close to the edge of object to be split, that is, external force pushes active contour to be transported towards target edges
It is dynamic, and internal force then keeps the slickness and topological of active contour, (corresponds to energy most at this time when snake point reaches equilbrium position
It is small), active contour will converge to the edge of object to be split.
Specifically, active contour is defined as geometrically being described as a two-dimensional space plane of delineation (x, y)
On a parametric curve：
R (s)=(x (s), y (s)) (4)
Wherein, the every bit r (s) on curve is known as snake point, its coordinate (x (s), y (s)) in the picture of each snake point
It indicates, these snake point straight lines or curve is connected and constitute entire profile.In this way, entire profile can be used only
These snake points indicate, and the deformation of active contour can also be completed by the movement of snake point.
The energy function of movable contour model completely defines its deformation behavior, the energy of each snake point in active contour
It is defined as Esnake：
Esnake=Eint+Eext (5)
Wherein, EintFor the internal energy at corresponding snake point, EextFor the external energy at corresponding snake point.
(it is minimum to correspond to energy at this time) when snake point reaches equilbrium position, active contour will converge to object to be split
Edge, realize image segmentation later.
Movable contour model namely snake models when carrying out image segmentation using snake algorithms, now illustrate segmentation
Effect：When being all made of snake algorithms and the equal iteration of algorithm 300 times, referring to cutting procedure schematic diagram shown in Fig. 6 A, Fig. 6 A are right
Side figure be treated in image to be split i.e. gray level image it is that cutting object is split as a result, not by object to be split whole
It splits；Referring to cutting procedure schematic diagram shown in Fig. 6 B, Fig. 6 B right part of flg is that cutting object progress is treated in entropy diagram picture
Segmentation as a result, it is possible to which object to be split is all split.As it can be seen that carrying out image segmentation, energy on the basis of entropy diagram picture
Enough obtain better segmentation effect.
More specifically, when carrying out image segmentation using snake algorithms, following partitioning scheme may be used：
A kind of partitioning scheme is the partitioning scheme searched from the outside to the core, and this partitioning scheme is suitble to the to be split right of big profile
As the noiseless object in outside of, the object to be split and the chaff interferent of inside is more.Another partitioning scheme is to search from inside to outside
Partitioning scheme, this partitioning scheme is suitble to the object to be split of little profile, the noiseless object in inside of the object to be split and outer
Portion's chaff interferent is more.
The present embodiment specifically may be used the movable contour model based on the field of forces GVF and realize image segmentation, GVF active contours
Partitioning algorithm has larger attractived region, can improve convergence of the deformable contour to boundary recess modification.
For example, with reference to search pattern schematic diagram from the outside to the core shown in Fig. 7 A, left side is object to be split, Ke Yi in figure
One initial active profile is set outside object to be split, the center of the initial active profile is the center of object to be split,
The gradual deformation of initial active profile finally converges to the edge of object to be split.
In another example referring to search pattern schematic diagram from inside to outside shown in Fig. 7 B, left side is object to be split in figure, can be with
One initial active profile is set inside object to be split, and the center of the initial active profile is in object to be split
The heart, the gradual deformation of initial active profile finally converge to the edge of object to be split.
In another example referring to two-way search pattern schematic diagram shown in Fig. 7 C, left side is object to be split, initial active in figure
The center of profile is not the center of object to be split, in this way, the top half of initial active profile is searched from the outside to the core, initial live
The lower half portion of driving wheel exterior feature is searched from inside to outside, and the gradual deformation of initial active profile finally converges to the edge of object to be split.
Cutting object is treated using GVF active contour partitioning algorithms and is split processing, it is advantageous that initial live driving wheel
Wide position can be random, and algorithm can scan for finally marking off accurate wheel according to the contour feature of area-of-interest
Wide position.
Embodiment three
Referring to Fig. 8, for a kind of flow diagram for image partition method that the present embodiment three provides.The image segmentation side
Method includes the following steps：
S801：Image to be split is obtained, the image to be split includes object to be split.
It should be noted that this step S801 is identical as the S101 in above-described embodiment one, related introduction refers to implementation
Example one, details are not described herein.
S802：Denoising is carried out to the image to be split by the way of shearing wave denoising, is obtained to be split after denoising
Image.
For the image to be split, during acquiring image data and during generating image, image
The random perturbation for generating each electronic device in equipment (such as medical imaging device), inevitably brings noise.Noise is made
It is a thorny problem in image processing process for the radio-frequency component in image, it can cause many image processing algorithms can not be just
Often execute or be not achieved ideal effect, and image segmentation is mainly according to the edge feature of object to be split, edge feature with make an uproar
Sound belongs to the radio-frequency component of image, so noise can bring image segmentation great interference, so, carrying out image segmentation
Before, it needs to carry out denoising to the image to be split.
However, it is generally the case that also result in the loss of marginal information while removal noise, in practical application, small echo
Denoising effect is preferable, about a kind of specific denoising method in Wavelet noise-eliminating method, i.e. shearing wave denoising method, the denoising method
Denoising can be carried out under the premise of reservation image detail ingredient as much as possible, to keep the edge of image special as possible
Sign.
S803：Using shock filter, edge enhancing is carried out to the image to be split after the denoising.
Filtering is equivalent to the process to image progress edge enhancing, deblurring.Shock filter is a kind of for scheming
The algorithm of image intensifying, theoretical foundation are hyperbolic equations theory and dullness rule, and shock filter may be used in the present embodiment, right
Object to be split in image to be split carries out edge prediction, and carries out image enhancement to image border, after image enhancement, can exist
Violent gray scale jump is formed at the origin-location of image border, to achieve the effect that image edge acuity, can be protected in this way
Stay the detailed information and location information of image border.
Specifically, the mathematic(al) representation of shock filter is：
Wherein, η is the direction of gradient image ▽ I；Sign () is sign function；| | | | it is euclideam norm；T is
Iterations；I0For observed image (initial value as iteration), the image to be split after the as described denoising.
Further, first derivation formula in above-mentioned formula (6) can also be used following formula (7) by the present embodiment
It replaces, improved shock filter (i.e. complex field shock filter-spreads shock filter model again) is obtained, using formula
(7) edge enhancing is carried out to the image to be split after the denoising, there will be more ideal image enhancement effects.
Wherein,For complex variableImaginary part；λ andRespectively represent multiple scalar (i.e. image I on the directions η two
The imaginary part of order derivative) and real scalar (i.e. the real part of second dervatives of the image I on the directions ξ)；IηηAnd IξξImage is indicated respectively
Second dervatives of the I on the directions η and ξ；θ is the angle close to 0；A is normal real number.
S804：The enhanced image to be split in edge is converted into entropy diagram picture, the picture of each pixel of the entropy diagram picture
Element value is the corresponding comentropy of the pixel.
S805：The unclosed region of object to be split in the entropy diagram picture is closed, obtains dividing preceding image.
It should be noted that step S804 and S805 respectively in above-described embodiment one S102 and S103 it is identical, it is related
Introduction refers to embodiment one, and details are not described herein.
S806：Gray scale stretching is carried out to image before the segmentation.
It should be noted that the preceding image of segmentation is carried out gray scale stretching, the edge of object to be split can be further increased
With the difference of image background, image segmentation is carried out on this basis, and segmentation effect will be more acurrate.
S807：The object to be split is partitioned into from image before the segmentation.
It should be noted that this step S807 is identical as the S104 in above-described embodiment one, related introduction refers to implementation
Example one, details are not described herein.
Example IV
Referring to Fig. 9, for a kind of composition schematic diagram for image segmentation device that the present embodiment four provides.The image segmentation device
900 include：
Image acquisition unit 901, for obtaining image to be split, the image to be split includes object to be split；
Image conversion unit 902, for the image to be split to be converted to entropy diagram picture, each pixel of the entropy diagram picture
The pixel value of point is the corresponding comentropy of the pixel；
Border seal unit 903, for sealing the unclosed region of the object to be split in the entropy diagram picture
It closes, obtains dividing preceding image；
Image segmentation unit 904, for being partitioned into the object to be split from image before the segmentation.
In a kind of embodiment of the application, described image converting unit 902 includes：
Patch frame selects subelement, for using each pixel in the image to be split as target pixel points,
Using the patch block of default size, the target pixel points are selected in the image center to be split；
Entropy computation subunit calculates the target picture for the gray value according to each pixel in the patch block
The corresponding comentropy of vegetarian refreshments；
Entropy judgment sub-unit, for judging whether each picture at least one pixel region and the pixel region
The corresponding comentropy of vegetarian refreshments is zero；
Entropy is reruned subelement, if for there are each pixels pair at least one pixel region and the pixel region
The comentropy answered is zero, then increases the size of the patch block, triggers the patch frame and selects subelement in the figure to be split
As center selects the target pixel points；
Image conversion subunit, if for there is no each pixels at least one pixel region and the pixel region
Corresponding comentropy is zero, then utilizes the corresponding comentropy formation entropy image of each pixel in the image to be split.
In a kind of embodiment of the application, the border seal unit 903 is specifically used for utilizing mathematical morphology
The unclosed region of the object to be split in the entropy diagram picture is closed in opening operation or closed operation.
In a kind of embodiment of the application, described image cutting unit 904 is specifically used for using based on active contour
The dividing method of model is partitioned into the object to be split from image before the segmentation.
In a kind of embodiment of the application, described device 900 further includes：
Image denoising unit is gone after obtaining image to be split in described image acquiring unit 901 using shearing wave
The mode made an uproar carries out denoising to the image to be split, obtains the image to be split after denoising.
In a kind of embodiment of the application, described device 900 further includes：
Edge enhancement unit, after obtaining the image to be split after denoising in described image denoising unit, using punching
Filter is hit, edge enhancing is carried out to the image to be split after the denoising.
In a kind of embodiment of the application, described device 900 further includes：
Gray scale stretching unit, after the image before the border seal unit 903 is divided, before the segmentation
Image carries out gray scale stretching.
Embodiment five
Referring to Figure 10, for a kind of hardware architecture diagram for image segmentation device that embodiment five provides, the system
1000 include memory 1001 and receiver 1002, and connect respectively with the memory 1001 and the receiver 1002
Processor 1003, the memory 1001 is for storing batch processing instruction, and the processor 1003 is for calling the storage
The program instruction that device 1001 stores executes following operation：
Image to be split is obtained, the image to be split includes object to be split；
The image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture is the pixel
The corresponding comentropy of point；
The unclosed region of the object to be split in the entropy diagram picture is closed, obtains dividing preceding image；
The object to be split is partitioned into from image before the segmentation.
In a kind of embodiment of the application, the processor 1003 is additionally operable to call the storage of the memory 1001
Program instruction executes following operation：
Using each pixel in the image to be split as target pixel points, the patch of default size is utilized
Block selects the target pixel points in the image center to be split；
According to the gray value of each pixel in the patch block, the corresponding comentropy of the target pixel points is calculated；
Judge whether that the corresponding comentropy of each pixel is equal at least one pixel region and the pixel region
It is zero；
If so, increase the size of the patch block, continues to execute and described select the mesh in the image center to be split
The step of marking pixel；
If it is not, then utilizing the corresponding comentropy formation entropy image of each pixel in the image to be split.
In a kind of embodiment of the application, the processor 1003 is additionally operable to call the storage of the memory 1001
Program instruction executes following operation：
Opening operation using mathematical morphology or closed operation, by the unclosed of the object to be split in the entropy diagram picture
It is closed in region.
In a kind of embodiment of the application, the processor 1003 is additionally operable to call the storage of the memory 1001
Program instruction executes following operation：
Using the dividing method based on movable contour model, it is described to be split right to be partitioned into from image before the segmentation
As.
In a kind of embodiment of the application, the processor 1003 is additionally operable to call the storage of the memory 1001
Program instruction executes following operation：
Denoising is carried out to the image to be split by the way of shearing wave denoising, obtains the image to be split after denoising.
In a kind of embodiment of the application, the processor 1003 is additionally operable to call the storage of the memory 1001
Program instruction executes following operation：
Using shock filter, edge enhancing is carried out to the image to be split after the denoising.
In a kind of embodiment of the application, the processor 1003 is additionally operable to call the storage of the memory 1001
Program instruction executes following operation：
Gray scale stretching is carried out to image before the segmentation.
In some embodiments, the processor 1003 can be central processing unit (Central Processing
Unit, CPU), the memory 1001 can be random access memory (Random Access Memory, RAM) type
Internal storage, the receiver 1002 can include General Physics interface, and the physical interface can be ether
(Ethernet) interface or asynchronous transfer mode (Asynchronous Transfer Mode, ATM) interface.The processor
1003, receiver 1002 and memory 1001 can be integrated into one or more independent circuits or hardware, such as：Special integrated electricity
Road (Application Specific Integrated Circuit, ASIC).
As seen through the above description of the embodiments, those skilled in the art can be understood that above-mentioned implementation
All or part of step in example method can add the mode of required general hardware platform to realize by software.Based on such
Understand, substantially the part that contributes to existing technology can be in the form of software products in other words for the technical solution of the application
It embodies, which can be stored in a storage medium, such as ROM/RAM, magnetic disc, CD, including several
Instruction is used so that a computer equipment (can be the network communications such as personal computer, server, or Media Gateway
Equipment, etc.) execute method described in certain parts of each embodiment of the application or embodiment.
It should be noted that each embodiment is described by the way of progressive in this specification, each embodiment emphasis is said
Bright is all difference from other examples, and just to refer each other for identical similar portion between each embodiment.For reality
For applying device disclosed in example, since it is corresponded to the methods disclosed in the examples, so description is fairly simple, related place
Referring to method part illustration.
It should also be noted that, herein, relational terms such as first and second and the like are used merely to one
Entity or operation are distinguished with another entity or operation, without necessarily requiring or implying between these entities or operation
There are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant are intended to contain
Lid non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also include other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
The foregoing description of the disclosed embodiments enables professional and technical personnel in the field to realize or use the application.
Various modifications to these embodiments will be apparent to those skilled in the art, as defined herein
General Principle can in other embodiments be realized in the case where not departing from spirit herein or range.Therefore, the application
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest range caused.
Claims (10)
1. a kind of image partition method, which is characterized in that including：
Image to be split is obtained, the image to be split includes object to be split；
The image to be split is converted into entropy diagram picture, the pixel value of each pixel of the entropy diagram picture is the pixel pair
The comentropy answered；
The unclosed region of the object to be split in the entropy diagram picture is closed, obtains dividing preceding image；
The object to be split is partitioned into from image before the segmentation.
2. according to the method described in claim 1, it is characterized in that, described be converted to entropy diagram picture by the image to be split, packet
It includes：
Using each pixel in the image to be split as target pixel points, using the patch block of default size,
The image center to be split selects the target pixel points；
According to the gray value of each pixel in the patch block, the corresponding comentropy of the target pixel points is calculated；
Judge whether that the corresponding comentropy of each pixel is zero at least one pixel region and the pixel region；
If so, increase the size of the patch block, continues to execute and described select the target picture in the image center to be split
The step of vegetarian refreshments；
If it is not, then utilizing the corresponding comentropy formation entropy image of each pixel in the image to be split.
3. according to the method described in claim 1, it is characterized in that, the object to be split by the entropy diagram picture
Unclosed region is closed, including：
Opening operation using mathematical morphology or closed operation, by the unclosed region of the object to be split in the entropy diagram picture
It is closed.
4. method according to any one of claims 1 to 3, which is characterized in that described from dividing in image before the segmentation
Go out the object to be split, including：
Using the dividing method based on movable contour model, the object to be split is partitioned into from image before the segmentation.
5. method according to any one of claims 1 to 3, which is characterized in that after the acquisition image to be split, packet
It includes：
Denoising is carried out to the image to be split by the way of shearing wave denoising, obtains the image to be split after denoising.
6. according to the method described in claim 5, it is characterized in that, after the image to be split obtained after denoising, also wrap
It includes：
Using shock filter, edge enhancing is carried out to the image to be split after the denoising.
7. method according to any one of claims 1 to 3, which is characterized in that it is described to obtain after dividing preceding image, also wrap
It includes：
Gray scale stretching is carried out to image before the segmentation.
8. a kind of image segmentation device, which is characterized in that including：
Image acquisition unit, for obtaining image to be split, the image to be split includes object to be split；
Image conversion unit, for the image to be split to be converted to entropy diagram picture, the picture of each pixel of the entropy diagram picture
Element value is the corresponding comentropy of the pixel；
Border seal unit is obtained for closing the unclosed region of the object to be split in the entropy diagram picture
Image before segmentation；
Image segmentation unit, for being partitioned into the object to be split from image before the segmentation.
9. device according to claim 8, which is characterized in that described image converting unit includes：
Patch frame selects subelement, for using each pixel in the image to be split as target pixel points, utilizing
The patch block of default size, the target pixel points are selected in the image center to be split；
Entropy computation subunit calculates the target pixel points for the gray value according to each pixel in the patch block
Corresponding comentropy；
Entropy judgment sub-unit, for judging whether each pixel at least one pixel region and the pixel region
Corresponding comentropy is zero；
Entropy is reruned subelement, if for there are each pixel at least one pixel region and the pixel region is corresponding
Comentropy is zero, then increases the size of the patch block, triggers the patch frame and selects subelement in the image to be split
Frame selects the target pixel points；
Image conversion subunit, if for there is no each pixels at least one pixel region and the pixel region to correspond to
Comentropy be zero, then utilize the image to be split in the corresponding comentropy formation entropy image of each pixel.
10. a kind of image segmentation device, which is characterized in that including：Processor, memory, system bus；
The processor and the memory are connected by the system bus；
The memory includes instruction for storing one or more programs, one or more of programs, and described instruction works as quilt
The processor makes the processor execute the method as described in any one of claim 1-7 when executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810005450.8A CN108305268B (en) | 2018-01-03 | 2018-01-03 | Image segmentation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810005450.8A CN108305268B (en) | 2018-01-03 | 2018-01-03 | Image segmentation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108305268A true CN108305268A (en) | 2018-07-20 |
| CN108305268B CN108305268B (en) | 2020-12-01 |
Family
ID=62868095
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810005450.8A Active CN108305268B (en) | 2018-01-03 | 2018-01-03 | Image segmentation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108305268B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110827301A (en) * | 2019-11-11 | 2020-02-21 | 京东数字科技控股有限公司 | Method and apparatus for processing image |
| CN113139936A (en) * | 2021-03-31 | 2021-07-20 | 武汉钢铁有限公司 | Image segmentation processing method and device |
| CN113191300A (en) * | 2020-10-14 | 2021-07-30 | 王建华 | Method, apparatus and computer readable medium for fingerprint image recognition |
| CN114359311A (en) * | 2022-03-17 | 2022-04-15 | 珠海市人民医院 | Medical image segmentation method and system based on artificial intelligence |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1885317A (en) * | 2006-07-06 | 2006-12-27 | 上海交通大学 | Adaptive edge detection method based on morphology and information entropy |
| US20120213440A1 (en) * | 2010-11-22 | 2012-08-23 | University Of Central Florida Research Foundation, Inc. | Systems and Methods for Automatically Identifying Shadows in Images |
| CN104637057A (en) * | 2015-02-04 | 2015-05-20 | 昆明理工大学 | Grayscale-gradient entropy multi-threshold fast division method based on genetic algorithm |
-
2018
- 2018-01-03 CN CN201810005450.8A patent/CN108305268B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1885317A (en) * | 2006-07-06 | 2006-12-27 | 上海交通大学 | Adaptive edge detection method based on morphology and information entropy |
| US20120213440A1 (en) * | 2010-11-22 | 2012-08-23 | University Of Central Florida Research Foundation, Inc. | Systems and Methods for Automatically Identifying Shadows in Images |
| CN104637057A (en) * | 2015-02-04 | 2015-05-20 | 昆明理工大学 | Grayscale-gradient entropy multi-threshold fast division method based on genetic algorithm |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110827301A (en) * | 2019-11-11 | 2020-02-21 | 京东数字科技控股有限公司 | Method and apparatus for processing image |
| CN110827301B (en) * | 2019-11-11 | 2023-09-26 | 京东科技控股股份有限公司 | Method and apparatus for processing image |
| CN113191300A (en) * | 2020-10-14 | 2021-07-30 | 王建华 | Method, apparatus and computer readable medium for fingerprint image recognition |
| CN113191300B (en) * | 2020-10-14 | 2024-01-26 | 神盾股份有限公司 | Method, apparatus and computer readable medium for fingerprint image recognition |
| CN113139936A (en) * | 2021-03-31 | 2021-07-20 | 武汉钢铁有限公司 | Image segmentation processing method and device |
| CN113139936B (en) * | 2021-03-31 | 2022-07-08 | 武汉钢铁有限公司 | Image segmentation processing method and device |
| CN114359311A (en) * | 2022-03-17 | 2022-04-15 | 珠海市人民医院 | Medical image segmentation method and system based on artificial intelligence |
| CN114359311B (en) * | 2022-03-17 | 2022-06-07 | 珠海市人民医院 | Medical image segmentation method and system based on artificial intelligence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108305268B (en) | 2020-12-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109949255B (en) | Image reconstruction method and device | |
| WO2022047625A1 (en) | Image processing method and system, and computer storage medium | |
| CN111798462B (en) | Automatic delineation method of nasopharyngeal carcinoma radiotherapy target area based on CT image | |
| CN110853051B (en) | Cerebrovascular image segmentation method based on multi-attention dense connection generation countermeasure network | |
| CN110276745B (en) | Pathological image detection algorithm based on generation countermeasure network | |
| CN108305268A (en) | A kind of image partition method and device | |
| JP2008511366A (en) | Feature-weighted medical object contour detection using distance coordinates | |
| CN113012172A (en) | AS-UNet-based medical image segmentation method and system | |
| CN112258514B (en) | Segmentation method of pulmonary blood vessels of CT (computed tomography) image | |
| Gauch et al. | The intensity axis of symmetry and its application to image segmentation | |
| CN113450396B (en) | Three-dimensional/two-dimensional image registration method and device based on bone characteristics | |
| CN110517198A (en) | High frequency sensitivity GAN network for LDCT image denoising | |
| Zhang et al. | A novel denoising method for CT images based on U-net and multi-attention | |
| CN110570394A (en) | medical image segmentation method, device, equipment and storage medium | |
| Du et al. | Blind image denoising via dynamic dual learning | |
| CN111383759A (en) | Automatic pneumonia diagnosis system | |
| CN116739899A (en) | Image super-resolution reconstruction method based on SAUGAN network | |
| CN107146211A (en) | Retinal vascular images noise-reduction method based on line spread function and bilateral filtering | |
| CN116894783A (en) | Metal artifact removal method for countermeasure generation network model based on time-varying constraint | |
| CN115909016A (en) | System, method, electronic device, and medium for analyzing fMRI image based on GCN | |
| Xu et al. | Dual-branch deep image prior for image denoising | |
| CN115100115A (en) | Organ image segmentation method, device, equipment and storage medium | |
| CN115205308A (en) | Fundus image blood vessel segmentation method based on linear filtering and deep learning | |
| CN110570417B (en) | Pulmonary nodule classification device and image processing equipment | |
| Sharma et al. | Brain region segmentation using low MSE based active contour model and convolutional neural network |
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 | ||
| CB02 | Change of applicant information |
Address after: 110179 No. 177-1 Innovation Road, Hunnan District, Shenyang City, Liaoning Province Applicant after: DongSoft Medical System Co., Ltd. Address before: 110179 No. 177-1 Innovation Road, Hunnan District, Shenyang City, Liaoning Province Applicant before: Dongruan Medical Systems Co., Ltd., Shenyang |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |