CN106651919A - Initialization apparatus and method of background image model, and image processing device - Google Patents
Initialization apparatus and method of background image model, and image processing device Download PDFInfo
- Publication number
- CN106651919A CN106651919A CN201510707124.8A CN201510707124A CN106651919A CN 106651919 A CN106651919 A CN 106651919A CN 201510707124 A CN201510707124 A CN 201510707124A CN 106651919 A CN106651919 A CN 106651919A
- Authority
- CN
- China
- Prior art keywords
- foreground
- background
- pixel
- model
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
Landscapes
- Image Analysis (AREA)
Abstract
The embodiments of the invention provide an initialization apparatus and method of a background image model, and an image processing device. The initialization method comprises the following steps: performing background image modeling at different time so as to obtain multiple background models; based on the multiple background models, obtaining corresponding multiple first foreground images; according to a statistical result about whether pixels in the multiple first foreground images are foreground pixels, generating a second foreground image; and according to the second foreground image, updating multiple background models so as to obtain an initial background image model. In such a way, the obtained initial background image model can be quite accurate, a ghosting artifact phenomenon in an image detection process is reduced, objects which have quite small moving speeds or are stationary in certain time can be detected, the accuracy of image detection is higher, and the capability of tolerance to noise is higher.
Description
Technical field
The present embodiments relate to technical field of image processing, more particularly to a kind of initial makeup of background image model
Put, method and image processing equipment.
Background technology
The extraction of background image is widely used in the fields such as picture control.For example in mobile object in detecting video,
The difference of present frame and reference frame can be compared, moving object is thus detected.Wherein, reference frame can be referred to as " the back of the body
Scape image ", or " background model " or " background image model " can also be used to be indicated.
There are certain methods to carry out the extraction of background model, such as frame difference method (Frame at present
Differencing), average filtration method (Mean filter) and background mixed model method (Background Mixture
Model).Initial background iconic model can be obtained using these methods, initial background iconic model is may then based on
It is updated, the background model after constantly being updated.
It should be noted that the introduction of technical background is intended merely to above it is convenient technical scheme is carried out it is clear,
Complete explanation, and facilitate the understanding of those skilled in the art and illustrate.Can not be merely because these schemes be at this
Bright background section is set forth and thinks that above-mentioned technical proposal is known to those skilled in the art.
The content of the invention
But, inventor has found:The initial background iconic model that background model initializing is obtained is carried out using current scheme
There are problems that inaccurate, ghost phenomenon often occurs in such as Jing, thus cause the background model after updating also inaccurate.
Embodiments provide a kind of apparatus for initializing of background image model, method and image processing equipment.
Want to obtain more accurately initial background iconic model.
A kind of one side according to embodiments of the present invention, there is provided apparatus for initializing of background image model, it is described first
Beginning makeup put including:
Background model acquiring unit, carries out background image and models to obtain multiple background models in different time;
First foreground image acquiring unit, based on the plurality of background model corresponding multiple first foreground images are obtained;
Whether the second foreground image signal generating unit, the pixel in the plurality of first foreground image is foreground pixel
Statistics is generating second foreground image;And
Initial model acquiring unit, updates the plurality of background model initial to obtain one according to second foreground image
Background image model.
A kind of second aspect according to embodiments of the present invention, there is provided initial method of background image model, it is described first
Beginning method includes:
Background image is carried out in different time to model to obtain multiple background models;
Corresponding multiple first foreground images are obtained based on the plurality of background model;
Whether the pixel in the plurality of first foreground image is the statistics of foreground pixel generating one
Two foreground images;And
Update the plurality of background model to obtain an initial background iconic model according to second foreground image.
3rd aspect according to embodiments of the present invention, there is provided a kind of image processing equipment, described image processing equipment bag
Include:The apparatus for initializing of background image model as above.
The beneficial effect of the embodiment of the present invention is:Corresponding multiple interim foreground pictures are obtained based on multiple background models
Picture;According to the pixel of relevant position in multiple interim foreground images it is whether foreground pixel generating a final foreground picture
Picture;And update multiple background models to obtain an initial background iconic model according to final foreground image.Thus, not only
More accurately initial background iconic model can be obtained, reduces the ghost phenomenon in image-detection process, and can
Detection translational speed is smaller or certain hour in static object, the accuracy of image detection is higher and tolerates noise
Ability it is higher.
With reference to explanation hereinafter and accompanying drawing, the particular implementation of the embodiment of the present invention is disclose in detail, specify this
The principle of bright embodiment can be in adopted mode.It should be understood that embodiments of the present invention in scope not thus
It is restricted.In the range of the spirit and terms of claims, embodiments of the present invention include many changes,
Modification and equivalent.
The feature for describing for a kind of embodiment and/or illustrating can be in same or similar mode one or more
It is combined with the feature in other embodiment used in individual other embodiment, or substitute in other embodiment
Feature.
It should be emphasized that term "comprises/comprising" refers to the presence of feature, one integral piece, step or component when using herein,
But it is not precluded from the presence of one or more further features, one integral piece, step or component or additional.
Description of the drawings
Included accompanying drawing is used for providing being further understood from the embodiment of the present invention, which constitutes of specification
Point, for illustrating embodiments of the present invention, and come together to explain the principle of the present invention with word description.Obviously
Ground, drawings in the following description are only some embodiments of the present invention, for those of ordinary skill in the art,
Without having to pay creative labor, can be with according to these other accompanying drawings of accompanying drawings acquisition.In the accompanying drawings:
Fig. 1 is a schematic diagram of the initial method of the background image model of the embodiment of the present invention 1;
Fig. 2 is the schematic diagram for how determining final foreground pixel of the embodiment of the present invention 1;
Fig. 3 is the schematic diagram being extended to foreground object region of the embodiment of the present invention 1;
Fig. 4 is another schematic diagram of the initial method of the background image model of the embodiment of the present invention 1;
Fig. 5 is a schematic diagram of the apparatus for initializing of the background image model of the embodiment of the present invention 2;
Fig. 6 is another schematic diagram of the apparatus for initializing of the background image model of the embodiment of the present invention 2;
Fig. 7 is a composition schematic diagram of the image processing equipment of the embodiment of the present invention 3.
Specific embodiment
Referring to the drawings, by description below, the aforementioned and further feature of the embodiment of the present invention will be apparent from.
In the specification and illustrated in the drawings, only certain exemplary embodiments of this invention is specifically disclosed, which show and wherein can adopt this
The some embodiments of the principle of bright embodiment, it will thus be appreciated that the invention is not restricted to described embodiment, phase
Instead, the embodiment of the present invention includes whole modifications, modification and the equivalent for falling within the scope of the appended claims.
In image-detection process, ghost phenomenon is likely to occur in following scene:For example, when a mobile object becomes
After (such as waiting the vehicle of traffic lights) for a period of time into stationary object, the mobile object may be considered as quiet
Only and be updated in background image or background model, ghost will be left when the object is moved again.Or,
When an object for having been considered to background image region (such as in parking lot stop over the automobile of several days) starts
When mobile, ghost phenomenon will be occurred when the object starts to move.
In the present embodiment, will illustrate by taking the picture control scene of field of traffic as an example.But the invention is not restricted to this,
In being also applied to other scenes.
Embodiment 1
The embodiment of the present invention provides a kind of initial method of background image model.Fig. 1 is the background of the embodiment of the present invention
One schematic diagram of the initial method of iconic model, as shown in figure 1, the initial method includes:
Step 101, carries out background image and models to obtain multiple background models in different time;
Step 102, based on multiple background models corresponding multiple first foreground images are obtained;
Whether step 103, the pixel in multiple first foreground images is the statistics of foreground pixel generating one
Individual second foreground image;
Step 104, updates multiple background models to obtain an initial background iconic model according to the second foreground image.
In the present embodiment, the video for including multiple two field pictures can be obtained by camera.The camera can be used
In the camera for carrying out traffic image monitoring, monitor area is shot incessantly;But the invention is not restricted to this,
Can also be other picture control scenes.
In the present embodiment, the background model in step 101 can be previously generated using prior art.For example
Can be in time T0Background model M is obtained according to a two field picture0, in time T1Background mould is obtained according to a two field picture
Type M1... ..., in time TiBackground model Mi is obtained according to a two field picture.The present embodiment can be according to multiple (examples
As N number of) background model to be obtaining a more accurately initial background iconic model.
In a step 102, corresponding N number of first foreground image can be obtained according to N number of background model (also may be used
Referred to as interim foreground image).With regard to how according to the corresponding foreground image of background model acquisition, it is possible to use existing skill
Any correlation technique in art.
In step 103, it is whether foreground pixel according to the pixel of relevant position in multiple first foreground images generating
One the second foreground image, specifically can include:Each location of pixels in for frame of video, judges the plurality of
Whether the pixel of relevant position is foreground pixel in one foreground image;To the individual of the corresponding foreground pixel of each location of pixels
Number is counted;Whether the pixel that each location of pixels is determined according to statistics is election (voted) foreground pixel
(alternatively referred to as final foreground pixel);And generate second foreground image using all of election foreground pixel
(alternatively referred to as final foreground image).
Wherein, determine whether the pixel of each location of pixels is election foreground pixel, specifically can be wrapped according to statistics
Include:The pixel of relevant position is the feelings of the number more than predetermined threshold value of foreground pixel in the plurality of first foreground image
Under condition, the pixel for determining the location of pixels is the election foreground pixel;And in the plurality of first foreground image
The pixel of middle relevant position is in the case that the number of foreground pixel is not more than predetermined threshold value, to determine the location of pixels
Pixel is not the election foreground pixel.
At step 104, N number of background model can be updated according to the second foreground image, is compared with obtaining one
Accurate initial background iconic model.With regard to how according to foreground image renewal background model, it is possible to use prior art
In any correlation technique.
Fig. 2 is the schematic diagram for how determining election foreground pixel (i.e. final foreground pixel) of the embodiment of the present invention,
As shown in Fig. 2 methods described includes:
Step 201, selects the location of pixels (x, y) in frame of video.
Step 202, selects a first foreground image Fj。
Step 203, judges the first foreground image FjWhether the pixel at middle location of pixels (x, y) place is foreground pixel;
If it is execution step 204, if not then execution step 205.
In the present embodiment, such as the first foreground image FjIt is binarized, if pixel in binary image
The pixel value of the pixel at position (x, y) place is 1, then can determine that the pixel is foreground pixel;If location of pixels
The pixel value of the pixel at (x, y) place is 0, then can determine that the pixel is not foreground pixel.But the invention is not restricted to
Whether this, can also adopt other methods to determine pixel for foreground pixel.
Step 204, counts to the number of the foreground pixel of location of pixels (x, y);
For example, would indicate that the individual numerical value of N of the number of foreground pixel(x, y)Plus 1.
Step 205, judges whether also exist without the first foreground image for processing;If it is execution step 202,
Next first foreground image is selected to proceed statistics;If do not had (i.e. all first foreground images have been disposed)
Then execution step 206.
Step 206, judges a numerical value of N(x, y)Whether predetermined threshold value is exceeded;If it exceeds predetermined threshold value then execution step
207;Execution step 208 if predetermined threshold value is not above.
In the present embodiment, predetermined threshold value can be for example N/2, but the invention is not restricted to this, can also be other values.
Step 207, the pixel at location of pixels (x, y) place is defined as to elect foreground pixel;
If exceeding the pixel at location of pixels (x, the y) place of the image of such as half in i.e. N number of first foreground image
For foreground pixel, then the pixel at location of pixels (x, y) place can be defined as electing foreground pixel (alternatively referred to as most
Whole foreground pixel).
Step 208, the pixel at location of pixels (x, y) place is not defined as electing foreground pixel;
For example, it is assumed that N is 10, predetermined threshold value is 5.For location of pixels (1,1), if counted at 10
In interim foreground image, (1,1) pixel at place is foreground pixel, then location of pixels in 8 interim foreground images
The corresponding pixel of location of pixels (1,1) is determined as into final foreground pixel;If counted in 10 interim prospects
In image, (1,1) pixel at place is foreground pixel to location of pixels, then by pixel in only 3 interim foreground images
The corresponding pixel in position (1,1) is defined as not being final foreground pixel.
Step 209, judges whether also exist without the location of pixels for processing;If it is execution step 201, select
Next location of pixels proceeds to process;The execution step if not having (i.e. all pixels position has been disposed)
210。
Step 210, using all of election foreground pixel the second foreground image is generated.
Schematically illustrated to how to obtain the second foreground image above.It should be noted that Fig. 1 and Fig. 2
The situation of the present invention is only diagrammatically illustrated, but the invention is not restricted to this.For example, it is also possible to be adjusted according to actual conditions
Order between each step, either increases and either reduces one of step or several steps.
In the present embodiment, N number of background model can respectively be carried out more according to the second foreground image in step 104
Newly.Wherein, for each background model, can be according to the foreground image areas in the second foreground image optionally more
New background model.
Specifically, can be using foreground image areas and adjacent area as prospect extended area;According to the current figure
Background model described in the area update beyond the prospect extended area in, and not by the prospect extended area
In updating the background model.
Fig. 3 is the schematic diagram being extended to foreground object region of the embodiment of the present invention, as shown in figure 3, can be with
Foreground image areas 302 in present image 300 except background image region 301 are extended.As shown in figure 3,
For example for certain pixel P on the edge of foreground image areas 302, by such as 5 pixels near pixel P
Also it is included into prospect extended area 303.
Thus, foreground image areas (such as moving object) and adjacent area will not be updated in background model,
The generation of ghost phenomenon can be reduced, and the smaller object moving of translational speed can be detected;The standard of image detection
Really property is higher and tolerates that the ability of noise is higher.
In the present embodiment, in the case of multiple background model identicals in the updated, one of background can be retained
Model and delete other background models obtaining the initial background iconic model;Multiple background models in the updated
In the case of differing, the background image can be re-started and modeled to obtain the plurality of background model.
Fig. 4 is another schematic diagram of the initial method of the background image model of the embodiment of the present invention, as shown in figure 4,
The initial method includes:
Step 401, models to obtain the background model of video background image is sometime carried out.
Step 402, judges whether to obtain N number of background model;If it is execution step 403;If without if after
Continuous execution step 401 obtains background model.
Step 403, based on N number of background model corresponding N number of first foreground image is obtained.
Step 404, for each location of pixels determines whether respective pixel is election foreground pixel.
Wherein it is possible to judge whether the respective pixel in N number of first foreground image is foreground pixel, and count every
The number of the foreground pixel of one location of pixels, and whether each pixel position is determined more than predetermined threshold value according to number
Put whether corresponding pixel is election foreground pixel.
Step 405, according to all of election foreground pixel the second foreground image is generated.
Step 406, according to second foreground image the plurality of background model is updated.
Wherein it is possible to using foreground image areas and adjacent area as prospect extended area;According to the present image
In the prospect extended area beyond area update described in background model, and not by the prospect extended area more
Newly in the background model.
Step 407, judges whether the multiple background models after updating are identical;Multiple background models in the updated are identical
In the case of execution step 408;Multiple background models in the updated differ in the case of execution step 403, can
To regain the plurality of background model, for example, the background image modeling is carried out again and obtains a new background mould
Type, then replaces a certain old background model to obtain the plurality of background model.
Step 408, retains one of background model and deletes other background models obtaining the initial background
Iconic model.
In the present embodiment, Fig. 1 or Fig. 4 show the initialization procedure of background model, the initial background image mould
Type can constantly be updated.For example according to background model M1, M2... ..., Mi is obtained more accurately just
Beginning background image model M'0Afterwards, it is also based on initial background iconic model M'0Constantly it is updated and obtains
M'1, M'2... ..., M'k;It is possible thereby to dynamically obtain more accurate background model.
From above-described embodiment, based on multiple background models corresponding multiple interim foreground images are obtained;According to multiple
Whether the pixel of relevant position is foreground pixel generating a final foreground image in interim foreground image;And according to institute
State final foreground image and update the plurality of background model to obtain an initial background iconic model.Thus, not only can be with
More accurately initial background iconic model is obtained, reduces the ghost phenomenon in image-detection process, and can be detected
Translational speed is smaller or certain hour in static object, the accuracy of image detection is higher and tolerates the energy of noise
Power is higher.
Embodiment 2
The embodiment of the present invention provides a kind of apparatus for initializing of background image model, and content same as Example 1 is no longer
Repeat.
Fig. 5 is a schematic diagram of the apparatus for initializing of the background image model of the embodiment of the present invention, as shown in figure 5, the back of the body
The apparatus for initializing 500 of scape iconic model includes:
Background model acquiring unit 501, carries out background image and models to obtain multiple background models in different time;
First foreground image acquiring unit 502, based on multiple background models corresponding multiple first foreground images are obtained;
Whether the second foreground image signal generating unit 503, the pixel in the plurality of first foreground image is prospect picture
The statistics of element is generating second foreground image;And
Initial model acquiring unit 504, updates the plurality of background model to obtain one according to second foreground image
Initial background iconic model.
Fig. 6 is another schematic diagram of the apparatus for initializing of the background image model of the embodiment of the present invention, as shown in fig. 6,
The apparatus for initializing 600 of background image model includes:Background model acquiring unit 501, the first foreground image obtain single
First 502, second foreground image signal generating unit 503 and initial model acquiring unit 504;As mentioned above.
As shown in fig. 6, the second foreground image signal generating unit 503 can include:
Pixel judging unit 601, for frame of video in each location of pixels, judge the plurality of first foreground image
Whether the pixel of middle relevant position is foreground pixel;
Pixels statisticses unit 602, counts to the number of the corresponding foreground pixel of each location of pixels;
As a result determining unit 603, whether the pixel that each location of pixels is determined according to statistics is election prospect
Pixel;And
Prospect signal generating unit 604, using all of election foreground pixel second foreground image is generated.
In the present embodiment, the result determining unit 603 specifically can be used for:In the plurality of first foreground image
The pixel of middle relevant position is in the case that the number of foreground pixel is more than predetermined threshold value, to determine the picture of the location of pixels
Element is the election foreground pixel;And the pixel of relevant position is foreground pixel in the plurality of first foreground image
Number be not more than predetermined threshold value in the case of, the pixel for determining the location of pixels is not the election foreground pixel.
As shown in fig. 6, the initial model acquiring unit 504 can include:
Context update unit 605, according to the subregion of second foreground image the plurality of back of the body is optionally updated
Scape model;And
Model treatment unit 606, in the case of the plurality of background model identical in the updated, retains one of them
Background model and delete other background models obtaining the initial background iconic model.
In the present embodiment, the context update unit 604 specifically can be used for:Not by second foreground image
Foreground object region and adjacent area update in the background model.Additionally, the model treatment unit 605
Can be also used for:In the case that the plurality of background model in the updated is differed, the plurality of background is regained
Model.
From above-described embodiment, based on multiple background models corresponding multiple interim foreground images are obtained;According to multiple
Whether the pixel of relevant position is foreground pixel generating a final foreground image in interim foreground image;And according to institute
State final foreground image and update the plurality of background model to obtain an initial background iconic model.Thus, not only can be with
More accurately initial background iconic model is obtained, reduces the ghost phenomenon in image-detection process, and can be detected
Translational speed is smaller or certain hour in static object, the accuracy of image detection is higher and tolerates the energy of noise
Power is higher.
Embodiment 3
The embodiment of the present invention provides a kind of image processing equipment, and described image processing equipment includes:As described in Example 2
Background image model apparatus for initializing.
Fig. 7 is a composition schematic diagram of the image processing equipment of the embodiment of the present invention.As shown in fig. 7, image procossing sets
Standby 700 can include:Central processing unit (CPU) 100 and memory 110;Memory 110 is coupled to centre
Reason device 100.Wherein the memory 110 can store various data;The program that additionally storage information is processed, and
The program is performed under the control of central processing unit 100.
In one embodiment, the function of the apparatus for initializing 500 or 600 of background image model can be integrated into
In central processing unit 100.Wherein, central processing unit 100 can be configured to realize background as described in Example 1
The initial method of iconic model.
In another embodiment, the apparatus for initializing 500 or 600 of background image model can be with central processing unit
Separate configuration, for example, can be configured to the apparatus for initializing 500 or 600 of background image model and central processing unit
The chip of 100 connections, the apparatus for initializing 500 of background image model is realized by the control of central processing unit 100
Or 600 function.
Additionally, as shown in fig. 7, image processing equipment 700 can also include:Input-output unit 120 and display are single
Unit 130 etc.;Wherein, similarly to the prior art, here is omitted for the function of above-mentioned part.It should be noted that
Image processing equipment 700 is also not necessary to include all parts shown in Fig. 7;Additionally, image processing equipment
700 can also may be referred to prior art including the part being not shown in Fig. 7.
The embodiment of the present invention also provides a kind of computer-readable program, wherein when performing the journey in image processing equipment
During sequence, described program causes computer that background image mould as described in Example 1 is performed in described image processing equipment
The initial method of type.
The embodiment of the present invention also provides a kind of storage medium of the computer-readable program that is stored with, wherein the computer can
Reader causes computer that the initialization of background image model as described in Example 1 is performed in image processing equipment
Method.
Apparatus and method more than of the invention can be realized by hardware, it is also possible to be realized by combination of hardware software.The present invention
It is related to such computer-readable program, when the program is performed by logical block, realizes can the logical block
Devices described above or component parts, or make the logical block realize various methods or step mentioned above.This
The bright storage medium further related to for storing procedure above, such as hard disk, disk, CD, DVD, flash memory
Deng.
Above in association with specific embodiment, invention has been described, it will be appreciated by those skilled in the art that this
A little descriptions are all exemplary, are not limiting the scope of the invention.Those skilled in the art can be according to this
Inventive principle makes various variants and modifications to the present invention, and these variants and modifications are also within the scope of the invention.
With regard to including the embodiment of above example, following note being also disclosed:
A kind of (note 1) apparatus for initializing of background image model, the apparatus for initializing includes:
Background model acquiring unit, carries out background image and models to obtain multiple background models in different time;
First foreground image acquiring unit, based on the plurality of background model corresponding multiple first foreground images are obtained;
Whether the second foreground image signal generating unit, the pixel in the plurality of first foreground image is foreground pixel
Statistics is generating second foreground image;And
Initial model acquiring unit, updates the plurality of background model initial to obtain one according to second foreground image
Background image model.
The apparatus for initializing of (note 2) according to note 1, wherein, the second foreground image signal generating unit bag
Include:
Pixel judging unit, for frame of video in each location of pixels, judge phase in the plurality of first foreground image
Whether the pixel for answering position is foreground pixel;
Pixels statisticses unit, counts to the number of the corresponding foreground pixel of each location of pixels;
As a result determining unit, whether the pixel that each location of pixels is determined according to statistics is election prospect picture
Element;And
Prospect signal generating unit, using all of election foreground pixel second foreground image is generated.
The apparatus for initializing of (note 3) according to note 2, wherein, the result determining unit is used for:Institute
The pixel for stating relevant position in multiple first foreground images is in the case that the number of foreground pixel is more than predetermined threshold value, really
The pixel of the fixed location of pixels is the election foreground pixel;And
The pixel of relevant position is that the number of foreground pixel is not more than predetermined threshold value in the plurality of first foreground image
In the case of, the pixel for determining the location of pixels is not the election foreground pixel.
The apparatus for initializing of (note 4) according to note 1, wherein, the initial model acquiring unit includes:
Context update unit, according to the subregion of second foreground image the plurality of background mould is optionally updated
Type;And
Model treatment unit, in the case of the plurality of background model identical in the updated, retains one of background
Model and delete other background models obtaining the initial background iconic model.
The apparatus for initializing of (note 5) according to note 4, wherein, the context update unit is used for:Will not
Foreground object region and adjacent area in second foreground image is updated in the background model.
The apparatus for initializing of (note 6) according to note 4, wherein, the model treatment unit is additionally operable to:
In the case that the plurality of background model after renewal is differed, the plurality of background model is regained.
A kind of (note 7) initial method of background image model, the initial method includes:
Background image is carried out in different time to model to obtain multiple background models;
Corresponding multiple first foreground images are obtained based on the plurality of background model;
Whether the pixel in the plurality of first foreground image is the statistics of foreground pixel generating one
Two foreground images;And
Update the plurality of background model to obtain an initial background iconic model according to second foreground image.
The initial method of (note 8) according to note 7, wherein, according in the plurality of first foreground image
Pixel be whether that the statistics of foreground pixel includes generating second foreground image:
Each location of pixels in for frame of video, judging the pixel of relevant position in the plurality of first foreground image is
It is no for foreground pixel;
The number of the corresponding foreground pixel of each location of pixels is counted;
Whether the pixel that each location of pixels is determined according to statistics is election foreground pixel;And
Second foreground image is generated using all of election foreground pixel.
The initial method of (note 9) according to note 8, wherein, each picture is determined according to statistics
Whether the pixel of plain position is that election foreground pixel includes:
The pixel of relevant position is the number of foreground pixel more than predetermined threshold value in the plurality of first foreground image
In the case of, the pixel for determining the location of pixels is the election foreground pixel;And
The pixel of relevant position is that the number of foreground pixel is not more than predetermined threshold value in the plurality of first foreground image
In the case of, the pixel for determining the location of pixels is not the election foreground pixel.
The initial method of (note 10) according to note 7, wherein, updated according to second foreground image
The plurality of background model is included with obtaining an initial background iconic model:
The plurality of background model is optionally updated according to the subregion of second foreground image;And
In the case of the plurality of background model identical in the updated, retain one of background model and delete other
Background model obtaining the initial background iconic model.
The initial method of (note 11) according to note 10, wherein, according to the portion of second foreground image
Subregion optionally updates the plurality of background model to be included:Not by the foreground object area in second foreground image
Domain and adjacent area are updated in the background model;
The initial method of (note 12) according to note 10, wherein, the initial method also includes:
In the case that the plurality of background model after renewal is differed, the plurality of background model is regained.
A kind of (note 13) image processing equipment, including:Background image mould as described in being attached 1 to 6 any one
The apparatus for initializing of type.
Claims (10)
1. a kind of apparatus for initializing of background image model, it is characterised in that the apparatus for initializing includes:
Background model acquiring unit, carries out background image and models to obtain multiple background models in different time;
First foreground image acquiring unit, based on the plurality of background model corresponding multiple first foreground images are obtained;
Whether the second foreground image signal generating unit, the pixel in the plurality of first foreground image is foreground pixel
Statistics is generating second foreground image;And
Initial model acquiring unit, updates the plurality of background model initial to obtain one according to second foreground image
Background image model.
2. apparatus for initializing according to claim 1, wherein, the second foreground image signal generating unit includes:
Pixel judging unit, for frame of video in each location of pixels, judge phase in the plurality of first foreground image
Whether the pixel for answering position is foreground pixel;
Pixels statisticses unit, counts to the number of the corresponding foreground pixel of each location of pixels;
As a result determining unit, whether the pixel that each location of pixels is determined according to statistics is election prospect picture
Element;And
Prospect signal generating unit, using all of election foreground pixel second foreground image is generated.
3. apparatus for initializing according to claim 2, wherein, the result determining unit is used for:Described
In multiple first foreground images the pixel of relevant position be the number of foreground pixel more than in the case of predetermined threshold value, it is determined that
The pixel of the location of pixels is the election foreground pixel;And
The pixel of relevant position is that the number of foreground pixel is not more than predetermined threshold value in the plurality of first foreground image
In the case of, the pixel for determining the location of pixels is not the election foreground pixel.
4. apparatus for initializing according to claim 1, wherein, the initial model acquiring unit includes:
Context update unit, according to the subregion of second foreground image the plurality of background mould is optionally updated
Type;And
Model treatment unit, in the case of the plurality of background model identical in the updated, retains one of background
Model and delete other background models obtaining the initial background iconic model.
5. apparatus for initializing according to claim 4, wherein, the context update unit is used for:Not by institute
State the foreground object region in the second foreground image and adjacent area is updated in the background model.
6. a kind of initial method of background image model, it is characterised in that the initial method includes:
Background image is carried out in different time to model to obtain multiple background models;
Corresponding multiple first foreground images are obtained based on the plurality of background model;
Whether the pixel in the plurality of first foreground image is the statistics of foreground pixel generating one
Two foreground images;And
Update the plurality of background model to obtain an initial background iconic model according to second foreground image.
7. initial method according to claim 6, wherein, according in the plurality of first foreground image
Whether pixel is that the statistics of foreground pixel includes generating second foreground image:
Each location of pixels in for frame of video, judging the pixel of relevant position in the plurality of first foreground image is
It is no for foreground pixel;
The number of the corresponding foreground pixel of each location of pixels is counted;
Whether the pixel that each location of pixels is determined according to statistics is election foreground pixel;And
Second foreground image is generated using all of election foreground pixel.
8. initial method according to claim 6, wherein, according to second foreground image updates
Multiple background models are included with obtaining an initial background iconic model:
The plurality of background model is optionally updated according to the subregion of second foreground image;Wherein not by institute
State the foreground object region in the second foreground image and adjacent area is updated in the background model;And
In the case of the plurality of background model identical in the updated, retain one of background model and delete other
Background model obtaining the initial background iconic model.
9. initial method according to claim 8, wherein, the initial method also includes:Updating
In the case that the plurality of background model afterwards is differed, the plurality of background model is regained.
10. a kind of image processing equipment, it is characterised in that described image processing equipment includes:Such as claim 1
The apparatus for initializing of described background image model.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510707124.8A CN106651919A (en) | 2015-10-27 | 2015-10-27 | Initialization apparatus and method of background image model, and image processing device |
JP2016207818A JP2017084365A (en) | 2015-10-27 | 2016-10-24 | Background image model initialization device and method, and image processing device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510707124.8A CN106651919A (en) | 2015-10-27 | 2015-10-27 | Initialization apparatus and method of background image model, and image processing device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106651919A true CN106651919A (en) | 2017-05-10 |
Family
ID=58714217
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510707124.8A Pending CN106651919A (en) | 2015-10-27 | 2015-10-27 | Initialization apparatus and method of background image model, and image processing device |
Country Status (2)
Country | Link |
---|---|
JP (1) | JP2017084365A (en) |
CN (1) | CN106651919A (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113311440B (en) * | 2020-02-26 | 2024-06-14 | 苏州佳世达电通有限公司 | Method for improving ultrasonic image quality and detection system thereof |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102314591A (en) * | 2010-07-09 | 2012-01-11 | 株式会社理光 | Method and equipment for detecting static foreground object |
CN102346854A (en) * | 2010-08-03 | 2012-02-08 | 株式会社理光 | Method and device for carrying out detection on foreground objects |
-
2015
- 2015-10-27 CN CN201510707124.8A patent/CN106651919A/en active Pending
-
2016
- 2016-10-24 JP JP2016207818A patent/JP2017084365A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102314591A (en) * | 2010-07-09 | 2012-01-11 | 株式会社理光 | Method and equipment for detecting static foreground object |
CN102346854A (en) * | 2010-08-03 | 2012-02-08 | 株式会社理光 | Method and device for carrying out detection on foreground objects |
Also Published As
Publication number | Publication date |
---|---|
JP2017084365A (en) | 2017-05-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111091091A (en) | Method, device and equipment for extracting target object re-identification features and storage medium | |
US8582915B2 (en) | Image enhancement for challenging lighting conditions | |
Wang et al. | Automatic natural video matting with depth | |
CN103186887B (en) | Image demister and image haze removal method | |
US20120019728A1 (en) | Dynamic Illumination Compensation For Background Subtraction | |
CN107005623B (en) | Method and apparatus for image processing | |
CN105631828A (en) | Image processing method and device | |
CN107295296B (en) | Method and system for selectively storing and recovering monitoring video | |
Liu et al. | Crowd counting with fully convolutional neural network | |
WO2017028047A1 (en) | Background model extracting method and apparatus and image processing device | |
CN113315884A (en) | Real-time video noise reduction method and device, terminal and storage medium | |
CN110114801A (en) | Display foreground detection device and method, electronic equipment | |
CN111161299A (en) | Image segmentation method, computer program, storage medium, and electronic device | |
Kryjak et al. | Real-time implementation of foreground object detection from a moving camera using the vibe algorithm | |
CN105869190A (en) | Background image reconstruction method, device and monitoring device | |
CN111192286A (en) | Image synthesis method, electronic device and storage medium | |
CN109905660B (en) | Method, device and computer readable storage medium for searching video event | |
JP2010257267A (en) | Device, method and program for detecting object area | |
CN110769262A (en) | Video image compression method and system | |
CN111160340B (en) | Moving object detection method and device, storage medium and terminal equipment | |
CN106651919A (en) | Initialization apparatus and method of background image model, and image processing device | |
WO2017028029A1 (en) | Background model extracting method and apparatus and image processing device | |
CN112084880A (en) | Image processing method, device, storage medium and equipment | |
JP5801614B2 (en) | Image processing apparatus and image processing method | |
CN110971811B (en) | Image screening method, system, terminal and computer readable storage medium |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170510 |