CN105096338B - Extracting of Moving Object and device - Google Patents
Extracting of Moving Object and device Download PDFInfo
- Publication number
- CN105096338B CN105096338B CN201410842471.7A CN201410842471A CN105096338B CN 105096338 B CN105096338 B CN 105096338B CN 201410842471 A CN201410842471 A CN 201410842471A CN 105096338 B CN105096338 B CN 105096338B
- Authority
- CN
- China
- Prior art keywords
- image
- displacement vector
- moving target
- region
- initialization
- 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.)
- Active
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of Extracting of Moving Object and device, this method includes：Two field pictures are read from continuous multiple frames image according to predetermined frame period, wherein, two field pictures are the first image and the second image；First image and the second image are divided into multiple subregions, multiple subregions carried out with the displacement vector that projection determines multiple subregions, global displacement vector of first image relative to the second image is determined according to the displacement vector of multiple subregions；It determines the ORB characteristic points of the first image and the second image, obtains the matching double points of two field pictures；The background dot of the first image and the second image is eliminated according to matching double points and global displacement vector, obtains the foreground point of the first image and the second image；The initialization partitioning boundary frame of watershed algorithm is determined according to the distribution of foreground point and movement destination image is extracted with this.By the present invention, solve the problems, such as that moving Object Segmentation is incomplete under dynamic background, the characteristic point in the characteristic point and background in target is distinguished.
Description
Technical field
The present invention relates to image processing field, in particular to a kind of Extracting of Moving Object and device.
Background technology
Moving object detection, extractive technique are the hot research problems of digital image processing field, in security protection, field of traffic
There is important application.Moving target recognition exactly comes out moving Object Segmentation present in video or image sequence.The relevant technologies
In, it can be achieved that the method for moving target recognition includes background subtraction method, frame differential method and optical flow method etc.；Common moving target
Velocity measuring system can be based on video camera, detection coil, UWB SAR, pulse clawback signal and realize.
Paper《Using the passive optical velocity measuring technique of sequence image》In recorded a kind of method, this method is using being installed on
Camera acquisition sequence image perpendicular to the ground immediately below unmanned plane, is matched by the SIFT feature between sequence image and completed
The measurement of unmanned plane speed, this paper only account for the situation that camera is disposed vertically, and the speed of measurement is the speed of unmanned plane itself
Degree, does not account for the tilted-putted situation of camera, without reference to the measurement of Moving Targets in Sequent Images speed.
In conclusion the processing method in the relevant technologies has at least the following problems：1) it is fixed for camera position
Situation is developed, and application range is restricted；2) to the moving target recognition under dynamic background, common background subtraction method is not
It can use, single moving object detection is easily multiple moving targets by frame differential method, and the problem of optical flow method is can not area
Partial objectives for light stream point and bias light flow point.
Invention content
The present invention provides a kind of Extracting of Moving Object and device, at least to solve prior art moving target recognition
The problem of inaccurate.
According to an aspect of the invention, there is provided a kind of Extracting of Moving Object, including：
Read two field pictures from continuous multiple frames image according to predetermined frame period, wherein, two field pictures for the first image and
Second image；
First image and the second image are divided into multiple subregions, multiple subregions carried out with the position that projection determines multiple subregions
Vector is moved, global displacement vector of first image relative to the second image is determined according to the displacement vector of multiple subregions；
It determines the ORB characteristic points of the first image and the second image, obtains the matching double points of two field pictures；
The background dot of the first image and the second image is eliminated according to matching double points and global displacement vector, obtains the first image
With the foreground point of the second image；
The initialization partitioning boundary frame of watershed algorithm is determined according to the distribution of foreground point；
Using watershed algorithm according to initialization partitioning boundary frame extraction movement destination image.
Further, multiple subregions include the first image and the global subregion of the second image entirety.
Further, the initialization partitioning boundary frame of watershed algorithm is determined according to the distribution of foreground point, including：
The point for selecting foreground point range image four direction frontier distance nearest, wherein, the nearest point of left and right frontier distance
Respectively A, B, the point nearest apart from upper and lower frontier distance is respectively C, D；
Respectively by 2 points of vertical lines for doing image up-and-down boundary of A, B, by 2 points of vertical lines for doing image right boundary of C, D,
Four directly cross to form rectangular area, wherein, the left and right width of rectangular area is a, upper-lower height b；
The left and right width in rectangular area is respectively extended into a/2, upper and lower each extension b/2 of height obtains initialization partitioning boundary frame.
Further, movement destination image is extracted according to initialization partitioning boundary frame using watershed algorithm, including：
Use watershed algorithm by the region segmentation initialized in partitioning boundary frame for multiple regions；
Determine that it is the region on moving target to meet first condition or the region of second condition in multiple regions, wherein, the
One condition includes foreground point for region and is not contacted with initialization partitioning boundary frame, and second condition is on region and moving target
Region contacts and is not contacted with initialization partitioning boundary frame；
The region merged on moving target obtains movement destination image.
Further, the above method further includes：
Determine the first boundary rectangle of the movement destination image of the first image, the second of the movement destination image of the second image
Boundary rectangle；
Determine the prospect match point that the first image is in the first boundary rectangle and the second image is in the second boundary rectangle
It is right；
The displacement vector of moving target is determined according to the displacement vector between prospect matching double points；
According to the displacement vector of moving target determined by continuous multiple frames image, moving target in continuous multiple frames image is determined
Displacement vector and global displacement vector absolute value of the difference maximum third image；
The speed of moving target is determined according to the state of third image, video camera and laser range finder.
According to another aspect of the present invention, a kind of moving target recognition device is provided, including：
Read module, for reading two field pictures from continuous multiple frames image according to predetermined frame period, wherein, two field pictures
For the first image and the second image；
Global displacement vector determination module, for the first image and the second image to be divided into multiple subregions, to multiple points
Area carries out the displacement vector that projection determines multiple subregions, determines the first image relative to second according to the displacement vector of multiple subregions
The global displacement vector of image；
Match point determining module for determining the ORB characteristic points of the first image and the second image, obtains of two field pictures
With point pair；
Foreground point determining module, for eliminating the first image and the second image according to matching double points and global displacement vector
Background dot obtains the foreground point of the first image and the second image；
Initialization module, for determining the initialization partitioning boundary frame of watershed algorithm according to the distribution of foreground point；
Moving target recognition module, using watershed algorithm according to initialization partitioning boundary frame extraction movement destination image.
Further, multiple subregions include the first image and the global subregion of the second image entirety.
Further, initialization module, including：
Selecting unit, for the point for selecting foreground point range image four direction frontier distance nearest, wherein, left and right side
The closest point in boundary is respectively A, B, and the point nearest apart from upper and lower frontier distance is respectively C, D；
For passing through 2 points of vertical lines for doing image up-and-down boundary of A, B respectively, an image left side is done at 2 points by C, D for processing unit
The vertical line of right margin, four directly cross to form rectangular area, wherein, the left and right width of rectangular area is a, upper-lower height b；
Expanding element, for the left and right width in rectangular area respectively to be extended a/2, upper and lower each extension b/2 of height is obtained initial
Change partitioning boundary frame.
Further, moving target recognition module, including：
Cutting unit, for using watershed algorithm by the region segmentation initialized in partitioning boundary frame for multiple regions；
Determination unit, the region for determining to meet first condition or second condition in multiple regions is on moving target
Region, wherein, first condition for region include foreground point and not with initialization partitioning boundary frame contact, second condition for region with
Region on moving target contacts and is not contacted with initialization partitioning boundary frame；
Combining unit, the region for merging on moving target obtain movement destination image.
Further, above device further includes：Movement velocity determining module, wherein, movement velocity determining module includes：
First determination unit, for determining the first boundary rectangle of the movement destination image of the first image, the second image
Second boundary rectangle of movement destination image；
Second determination unit, for determining the first image in the first boundary rectangle and the second image is in the second external square
Prospect matching double points in shape；
Third determination unit, for determining that the displacement of moving target is sweared according to the displacement vector between prospect matching double points
Amount；
4th determination unit for the displacement vector of the moving target according to determined by continuous multiple frames image, determines continuous
The third image of the displacement vector of moving target and the absolute value of the difference maximum of global displacement vector in multiple image；
5th determination unit determines the speed of moving target according to the state of third image, video camera and laser range finder.
By the present invention, remote moving target recognition and tachometric survey, foreground features point under dynamic background are realized
With the separation of background characteristics point, the complete and automatic segmentation of moving target, global displacement vector by target local displacement shadow
Sound is small.
Description of the drawings
Attached drawing described herein is used to provide further understanding of the present invention, and forms the part of the application, this hair
Bright illustrative embodiments and their description do not constitute improper limitations of the present invention for explaining the present invention.In the accompanying drawings：
Fig. 1 is the flow chart of Extracting of Moving Object according to embodiments of the present invention；
Fig. 2 is the structure diagram of moving target recognition device according to embodiments of the present invention；
Fig. 3 is the flow chart that optional velocity to moving target determines method according to embodiments of the present invention；
Fig. 4 is the schematic diagram one of optional picture portion according to embodiments of the present invention；
Fig. 5 is the schematic diagram two of optional picture portion according to embodiments of the present invention.
Specific embodiment
Come that the present invention will be described in detail below with reference to attached drawing and in conjunction with the embodiments.It should be noted that do not conflicting
In the case of, the feature in embodiment and embodiment in the application can be combined with each other.
A kind of Extracting of Moving Object is provided in the present embodiment, and Fig. 1 is movement mesh according to embodiments of the present invention
The flow chart of extracting method is marked, as shown in Figure 1, the flow includes the following steps：
Step S102 reads two field pictures according to predetermined frame period from continuous multiple frames image, wherein, two field pictures
One image and the second image；
Optionally, above-mentioned predetermined frame period is 1, and the first image is the image before the second image.
First image and the second image are divided into multiple subregions by step S104, and it is determining more that multiple subregions are carried out with projection
The displacement vector of a subregion determines that the first image is sweared relative to the global displacement of the second image according to the displacement vector of multiple subregions
Amount；
Optionally, the whole whole subregion as a subregion of the partial-partition of subregion including image section and image, from
And it can realize the combination of subregion and the overall situation.
Step S106 determines the ORB characteristic points of the first image and the second image, obtains the matching double points of two field pictures；
Step S108 eliminates the background dot of the first image and the second image according to matching double points and global displacement vector, obtains
To the foreground point of the first image and the second image；
Step S110 determines the initialization partitioning boundary frame of watershed algorithm according to the distribution of foreground point；
Step S112, using watershed algorithm according to initialization partitioning boundary frame extraction movement destination image.
In an optional embodiment of the embodiment of the present invention, above-mentioned steps S110 includes：
A selects the nearest point of foreground point range image four direction frontier distance, wherein, left and right frontier distance is nearest
Point is respectively A, B, and the point nearest apart from upper and lower frontier distance is respectively C, D；
Respectively by 2 points of vertical lines for doing image up-and-down boundary of A, B, hanging down for image right boundary is done at 2 points by C, D by b
Line, four directly cross to form rectangular area, wherein, the left and right width of rectangular area is a, upper-lower height b；
The left and right width in rectangular area is respectively extended a/2 by c, and upper and lower each extension b/2 of height obtains initialization partitioning boundary
Frame.
In an optional embodiment of the embodiment of the present invention, above-mentioned steps S112, using watershed algorithm according to just
Beginningization partitioning boundary frame extracts movement destination image, including：
1, use watershed algorithm by the region segmentation initialized in partitioning boundary frame for multiple regions；
2, determine that it is the region on moving target to meet first condition or the region of second condition in multiple regions, wherein,
First condition includes foreground point for region and is not contacted with initialization partitioning boundary frame, and second condition is on region and moving target
Region contact and not with initialization partitioning boundary frame contact；
3, the region merged on moving target obtains movement destination image.
In an optional embodiment of the embodiment of the present invention, it is also based on the speed that the above method determines moving target
Degree, the above method further includes thus：
Determine the first boundary rectangle of the movement destination image of the first image, the second of the movement destination image of the second image
Boundary rectangle；
Determine the prospect match point that the first image is in the first boundary rectangle and the second image is in the second boundary rectangle
It is right；
The displacement vector of moving target is determined according to the displacement vector between prospect matching double points；
According to the displacement vector of moving target determined by continuous multiple frames image, moving target in continuous multiple frames image is determined
Displacement vector and global displacement vector absolute value of the difference maximum third image；
The speed of moving target is determined according to the state of third image, video camera and laser range finder.
Additionally provide a kind of moving target recognition device in the present embodiment, which is used to implement above-described embodiment and excellent
Embodiment is selected, had carried out repeating no more for explanation.As used below, term " module " can realize predetermined function
Software and/or hardware combination.Although following embodiment described device is preferably realized with software, hardware,
Or the realization of the combination of software and hardware is also what may and be contemplated.
Fig. 2 is the structure diagram of moving target recognition device according to embodiments of the present invention, as shown in Fig. 2, the device packet
It includes：
Read module 10, for reading two field pictures from continuous multiple frames image according to predetermined frame period, wherein, two frame figures
As being the first image and the second image；
Global displacement vector determination module 20 is connected with read module 10, for the first image and the second image to be divided
For multiple subregions, multiple subregions are carried out with the displacement vector that projection determines multiple subregions, it is true according to the displacement vector of multiple subregions
Determine global displacement vector of first image relative to the second image；
Match point determining module 30 is connected with global displacement vector determination module 20, for determining the first image and second
The ORB characteristic points of image, obtain the matching double points of two field pictures；
Foreground point determining module 40 is connected with match point determining module 30, for according to matching double points and global displacement arrow
Amount eliminates the background dot of the first image and the second image, obtains the foreground point of the first image and the second image；
Initialization module 50 is connected with foreground point determining module 40, for determining that watershed is calculated according to the distribution of foreground point
The initialization partitioning boundary frame of method；
Moving target recognition module 60 is connected with initialization module 50, using watershed algorithm according to initialization segmentation side
Boundary's frame extraction movement destination image.
In an optional embodiment of the embodiment of the present invention, above-mentioned multiple subregions include the first image and the second figure
As whole global subregion.
In an optional embodiment of the embodiment of the present invention, initialization module 50 includes：
Selecting unit, for the point for selecting foreground point range image four direction frontier distance nearest, wherein, left and right side
The closest point in boundary is respectively A, B, and the point nearest apart from upper and lower frontier distance is respectively C, D；
For passing through 2 points of vertical lines for doing image up-and-down boundary of A, B respectively, an image left side is done at 2 points by C, D for processing unit
The vertical line of right margin, four directly cross to form rectangular area, wherein, the left and right width of rectangular area is a, upper-lower height b；
Expanding element, for the left and right width in rectangular area respectively to be extended a/2, upper and lower each extension b/2 of height is obtained initial
Change partitioning boundary frame.
In an optional embodiment of the embodiment of the present invention, moving target recognition module 60 includes：
Cutting unit, for using watershed algorithm by the region segmentation initialized in partitioning boundary frame for multiple regions；
Determination unit, the region for determining to meet first condition or second condition in multiple regions is on moving target
Region, wherein, first condition for region include foreground point and not with initialization partitioning boundary frame contact, second condition for region with
Region on moving target contacts and is not contacted with initialization partitioning boundary frame；
Combining unit, the region for merging on moving target obtain movement destination image.
In an optional embodiment of the embodiment of the present invention, above device further includes：Movement velocity determining module,
In, movement velocity determining module includes：
First determination unit, for determining the first boundary rectangle of the movement destination image of the first image, the second image
Second boundary rectangle of movement destination image；
Second determination unit, for determining the first image in the first boundary rectangle and the second image is in the second external square
Prospect matching double points in shape；
Third determination unit, for determining that the displacement of moving target is sweared according to the displacement vector between prospect matching double points
Amount；
4th determination unit for the displacement vector of the moving target according to determined by continuous multiple frames image, determines continuous
The third image of the displacement vector of moving target and the absolute value of the difference maximum of global displacement vector in multiple image；
5th determination unit determines the speed of moving target according to the state of third image, video camera and laser range finder.
Optional embodiment
In the optional embodiment, moving target recognition and speed measurement method under a kind of dynamic background are provided,
When aerial platform (manned helicopter, unmanned plane helicopter, dirigible when) hovering when by sequence video frame or being continuously shot
Image is managed, and to extract moving target, and calculates the speed of moving target, as shown in figure 3, this method specifically includes following step
Suddenly：
Step S302 stores image.The m frame images that video camera or camera are continuously shot are sequentially stored into f1(x, y), f2
(x, y), f3(x, y) ... fm(x, y)：As input present frame or image f0When (x, y), successively by fm-1(x, y) is stored in fm(x,
Y), fm-2(x, y) is stored in fm-1(x, y) ..., f1(x, y) is stored in f2(x, y), f0(x, y) is stored in f1(x, y), by f1(x, y), f3
In deposit the image PicA, PicB of (x, y), the shoot with video-corder period or the shooting period of camera of video frame are T.
In the optional embodiment, the global displacement Vector operation side being combined is projected with global using subregion projection
Method can reduce the influence of local moving objects, and obtained global displacement vector is more accurate.
Step S304, ORB feature point extraction.PicA is calculated, the ORB characteristic points of PicB obtain the match point of two field pictures
It is right.
Step S306, global motion vector estimation.Global displacement vectors of the PicA relative to PicB is calculated using sciagraphy,
Assuming that picture traverse is w, it is highly h, the upper left corner is image origin, downwards and as the right side is the just position upper left of this subregion
Angle starting point x coordinate, upper left corner starting point y-coordinate, rectangle width, rectangular elevation expression, as shown in Figures 4 and 5, each subregion A1、A2、A3、
A4、A5、A6It is represented by：
Each subregion projection carries out the displacement vector that correlation computations obtain and is respectively： (a, b, c ∈ 1,2,3,4,5 and a ≠ b, b
≠ c, a ≠ c) (1)
When above formula works as SminGlobal displacement vector when obtaining minimum value：
Step S308, foreground point extraction.
According to the global displacement vector that the matched ORB characteristic points of step S304 and step S306 are calculated eliminate image PicA and
PicB background dots, obtain foreground point.It is respectively P to match point that image PicA and PicB, which have n,A1、PA2……PAn-1、PAnAnd PB1、
PB2……PBn-1、PBnDisplacement vector be respectively
Prospect match point meets following relationship：
When
Foreground features point is determined by the way of ORB Feature Points Matching combination global displacement vectors, and arithmetic speed is fast.
Step S310 initializes partitioning boundary frame.
Foreground point (point on the moving target) distribution acquired according to step S308, acquires the initialization point of watershed algorithm
Bounding box is cut, method is as follows：
The nearest point of selection foreground point range image four direction frontier distance respectively, the nearest point of left and right frontier distance
Respectively A, B, the point nearest apart from upper and lower frontier distance are respectively C, D, respectively by A, B 2 points do image up-and-down boundary
Vertical line, by 2 points of vertical lines for doing image right boundary of C, D, four rectangular areas directly to cross are denoted as RectA, and left and right is wide
It spends for a, upper-lower height b；The left and right width of RectA is respectively extended into a/2, upper and lower each extension b/2 of height obtains rectangular area
RectB。
In the optional embodiment, the initialization partitioning boundary of watershed algorithm is given using completely automatic mode
Frame can avoid cumbersome manual operation.
Step S312, moving target recognition.
Ask for the outer boundary frame of target, RectB be divided by several regions using watershed algorithm, some region meet with
A), b) first condition thinks that the region is the region on moving target：
A), the region includes foreground point and is not in contact with the bounding box of RectB；
B), the region contacts with the region on moving target and is not contacted with the bounding box of RectB.
Image is the moving target that detects after region merging technique on moving target is divided, ask for PicA and
The boundary rectangle of PicB moving targets is denoted as RectCA and RectCB respectively, and the centre coordinate for remembering RectCA is PointCA coordinates
It is worth for (CAx, CAy), width is WidthCA pixels, is highly HeightCA pixels, the centre coordinate for remembering RectCB is
PointCB coordinate values are (CBx, CBy), and width is WidthCB pixels, is highly HeightCB, the moving target of being detected
Minimum dimension is RectS pixels, thinks to detect moving target when meeting following relationship：
By the step, can complete parttion go out moving target, it is several targets to avoid single moving Object Segmentation
Situation.
Step S314, displacement of targets Vector operation.
Prospect matching double points quantity of the statistical picture PicA and PicB in region RectCA, RectCB is denoted as PointM,
Displacement vector between match point is followed successively byMoving target vector is
In this step, velocity to moving target is calculated using the ORB characteristic points in target, multiple spot is calculated to reduce to calculate and be missed
Difference.
Step S316, velocity to moving target calculate.
Successively by f3(x, y), f4(x, y) ... fkIn (x, y) deposit image PicB, according to step 2) -7) process calculate position
Vector (unit pixel) is moved,
Work as ScMaximum value SmaxWhen corresponding deposit PicB images be fz(x, y), then the speed calculation method of moving target is such as
Under, f1(x, y) focal length of camera is f1, laser range finder ranging is l1, fn(x, y) focal length of camera is fy, laser range finder survey
Away from for ly, minister video camera pixel dimension is that the velocity to moving target that μ is measured is：
Velocity to moving target can be calculated according to the state and measured value of video camera and laser range finder, can adapt to aerial
It remote observation and tests the speed.
The optional embodiment solves the problems, such as that moving Object Segmentation is incomplete under usual dynamic background, and by target
On characteristic point and background on characteristic point distinguished.The velocity to moving target for solving the problems, such as mobile camera measures,
It can be applied to aerial platform and realize testing the speed at a distance to moving target.
Obviously, those skilled in the art should be understood that each module of the above-mentioned present invention or each step can be with general
Computing device realize that they can concentrate on single computing device or be distributed in multiple computing devices and be formed
Network on, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
It is performed in the storage device by computing device, and in some cases, it can be to be different from shown in sequence herein performs
The step of going out or describing they are either fabricated to each integrated circuit modules respectively or by multiple modules in them or
Step is fabricated to single integrated circuit module to realize.It to be combined in this way, the present invention is not limited to any specific hardware and softwares.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, that is made any repaiies
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of Extracting of Moving Object, which is characterized in that including：
Read two field pictures from continuous multiple frames image according to predetermined frame period, wherein, the two field pictures for the first image and
Second image；
Described first image and second image are divided into multiple subregions, it is determining described that projection is carried out to the multiple subregion
The displacement vector of multiple subregions determines described first image relative to second figure according to the displacement vector of the multiple subregion
The global displacement vector of picture；
It determines described first image and the ORB characteristic points of second image, obtains the matching double points of two field pictures；
The background dot of described first image and second image is eliminated according to the matching double points and the global displacement vector,
Obtain the foreground point of described first image and second image；
The initialization partitioning boundary frame of watershed algorithm is determined according to the distribution of the foreground point；
Using watershed algorithm according to the initialization partitioning boundary frame extraction movement destination image.
2. according to the method described in claim 1, it is characterized in that, the multiple subregion includes described first image and described
The global subregion of second image entirety.
3. according to the method described in claim 1, it is characterized in that, the distribution according to the foreground point determines watershed algorithm
Partitioning boundary frame is initialized, including：
The point for selecting foreground point range image four direction frontier distance nearest, wherein, the nearest point difference of left and right frontier distance
For A, B, the point nearest apart from upper and lower frontier distance is respectively C, D；
Respectively by 2 points of vertical lines for doing image up-and-down boundary of A, B, by 2 points of vertical lines for doing image right boundary of C, D, four
Directly cross to form rectangular area, wherein, the left and right width of the rectangular area is a, upper-lower height b；
The left and right width in the rectangular area is respectively extended into a/2, upper and lower each extension b/2 of height obtains initialization partitioning boundary frame.
4. the method according to claim 1 or 3, which is characterized in that divided using watershed algorithm according to the initialization
Bounding box extracts movement destination image, including：
Use watershed algorithm by the region segmentation in the initialization partitioning boundary frame for multiple regions；
Determine that it is the region on moving target to meet first condition or the region of second condition in the multiple region, wherein, institute
State first condition for region include foreground point and not with it is described initialization partitioning boundary frame contact, the second condition for region with
Region on moving target contacts and is not contacted with the initialization partitioning boundary frame；
The region merged on the moving target obtains movement destination image.
5. according to the method in any one of claims 1 to 3, which is characterized in that further include：
Determine the first boundary rectangle of the movement destination image of described first image, the movement destination image of second image
Second boundary rectangle；
Described first image is determined in first boundary rectangle and second image is in second boundary rectangle
Prospect matching double points；
The displacement vector of moving target is determined according to the displacement vector between prospect matching double points；
According to the displacement vector of moving target determined by the continuous multiple frames image, determine to move in the continuous multiple frames image
The third image of the displacement vector of target and the absolute value of the difference maximum of global displacement vector；
The speed of moving target is determined according to the state of the third image, video camera and laser range finder.
6. a kind of moving target recognition device, which is characterized in that including：
Read module, for reading two field pictures from continuous multiple frames image according to predetermined frame period, wherein, the two field pictures
For the first image and the second image；
Global displacement vector determination module, for described first image and second image to be divided into multiple subregions, to institute
It states multiple subregions and carries out the displacement vector that projection determines the multiple subregion, institute is determined according to the displacement vector of the multiple subregion
State global displacement vector of first image relative to second image；
Match point determining module for determining the ORB characteristic points of described first image and second image, obtains two field pictures
Matching double points；
Foreground point determining module, for eliminating described first image and institute according to the matching double points and the global displacement vector
The background dot of the second image is stated, obtains the foreground point of described first image and second image；
Initialization module, for determining the initialization partitioning boundary frame of watershed algorithm according to the distribution of the foreground point；
Moving target recognition module, using watershed algorithm according to the initialization partitioning boundary frame extraction movement destination image.
7. device according to claim 6, which is characterized in that the multiple subregion includes described first image and described
The global subregion of second image entirety.
8. device according to claim 6, which is characterized in that the initialization module, including：
Selecting unit, for the point for selecting foreground point range image four direction frontier distance nearest, wherein, left and right boundary away from
It is respectively A, B from nearest point, the point nearest apart from upper and lower frontier distance is respectively C, D；
For passing through 2 points of vertical lines for doing image up-and-down boundary of A, B respectively, image left and right side is done at 2 points by C, D for processing unit
The vertical line on boundary, four directly cross to form rectangular area, wherein, the left and right width of the rectangular area is a, upper-lower height b；
Expanding element, for the left and right width in the rectangular area respectively to be extended a/2, upper and lower each extension b/2 of height is obtained initial
Change partitioning boundary frame.
9. the device according to claim 6 or 8, which is characterized in that the moving target recognition module, including：
Cutting unit, for using watershed algorithm by the region segmentation in the initialization partitioning boundary frame for multiple regions；
Determination unit, the region for determining to meet first condition or second condition in the multiple region is on moving target
Region, wherein, the first condition includes foreground point for region and is not contacted with the initialization partitioning boundary frame, described second
Condition is contacted with the region on moving target for region and is not contacted with the initialization partitioning boundary frame；
Combining unit obtains movement destination image for merging the region on the moving target.
10. the device according to any one of claim 6 to 8, which is characterized in that further include：Movement velocity determining module,
Wherein, the movement velocity determining module includes：
First determination unit, for determining the first boundary rectangle of the movement destination image of described first image, second figure
Second boundary rectangle of the movement destination image of picture；
Second determination unit, for determining described first image in first boundary rectangle and second image is in institute
State the prospect matching double points in the second boundary rectangle；
Third determination unit, for determining the displacement vector of moving target according to the displacement vector between prospect matching double points；
4th determination unit for the displacement vector of the moving target according to determined by the continuous multiple frames image, determines described
The third image of the displacement vector of moving target and the absolute value of the difference maximum of global displacement vector in continuous multiple frames image；
5th determination unit determines the speed of moving target according to the state of the third image, video camera and laser range finder.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410842471.7A CN105096338B (en) | 2014-12-30 | 2014-12-30 | Extracting of Moving Object and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410842471.7A CN105096338B (en) | 2014-12-30 | 2014-12-30 | Extracting of Moving Object and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105096338A CN105096338A (en) | 2015-11-25 |
| CN105096338B true CN105096338B (en) | 2018-06-22 |
Family
ID=54576677
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410842471.7A Active CN105096338B (en) | 2014-12-30 | 2014-12-30 | Extracting of Moving Object and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105096338B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020042156A1 (en) * | 2018-08-31 | 2020-03-05 | 深圳市道通智能航空技术有限公司 | Motion area detection method and device, and unmanned aerial vehicle |
| CN109460764B (en) * | 2018-11-08 | 2022-02-18 | 中南大学 | Satellite video ship monitoring method combining brightness characteristics and improved interframe difference method |
| TWI790508B (en) * | 2020-11-30 | 2023-01-21 | 宏碁股份有限公司 | Blood vessel detecting apparatus and blood vessel detecting method based on image |
| CN115797164B (en) * | 2021-09-09 | 2023-12-12 | 同方威视技术股份有限公司 | Image stitching method, device and system in fixed view field |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102098440A (en) * | 2010-12-16 | 2011-06-15 | 北京交通大学 | Electronic image stabilizing method and electronic image stabilizing system aiming at moving object detection under camera shake |
| CN103065126A (en) * | 2012-12-30 | 2013-04-24 | 信帧电子技术（北京）有限公司 | Re-identification method of different scenes on human body images |
| CN103971114A (en) * | 2014-04-23 | 2014-08-06 | 天津航天中为数据系统科技有限公司 | Forest fire detection method based on aerial remote sensing |
| CN104200487A (en) * | 2014-08-01 | 2014-12-10 | 广州中大数字家庭工程技术研究中心有限公司 | Target tracking method based on ORB characteristics point matching |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5743142B2 (en) * | 2011-03-29 | 2015-07-01 | ソニー株式会社 | Image processing apparatus and method, and program |
-
2014
- 2014-12-30 CN CN201410842471.7A patent/CN105096338B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102098440A (en) * | 2010-12-16 | 2011-06-15 | 北京交通大学 | Electronic image stabilizing method and electronic image stabilizing system aiming at moving object detection under camera shake |
| CN103065126A (en) * | 2012-12-30 | 2013-04-24 | 信帧电子技术（北京）有限公司 | Re-identification method of different scenes on human body images |
| CN103971114A (en) * | 2014-04-23 | 2014-08-06 | 天津航天中为数据系统科技有限公司 | Forest fire detection method based on aerial remote sensing |
| CN104200487A (en) * | 2014-08-01 | 2014-12-10 | 广州中大数字家庭工程技术研究中心有限公司 | Target tracking method based on ORB characteristics point matching |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105096338A (en) | 2015-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110221603B (en) | Remote obstacle detection method based on laser radar multi-frame point cloud fusion | |
| US9274204B2 (en) | Camera tracing and surveillance system and method for security using thermal image coordinate | |
| KR100422370B1 (en) | An Apparatus and Method to Measuring Dimensions of 3D Object on a Moving Conveyor | |
| CN105096338B (en) | Extracting of Moving Object and device | |
| JP6524529B2 (en) | Building limit judging device | |
| CN105975923B (en) | Method and system for tracking human objects | |
| US11132802B2 (en) | Method of detecting moving objects from a temporal sequence of images | |
| JP2019185281A5 (en) | ||
| KR20150029551A (en) | Determining source lane of moving item merging into destination lane | |
| Li et al. | Weak moving object detection in optical remote sensing video with motion-drive fusion network | |
| CN112613568B (en) | Target identification method and device based on visible light and infrared multispectral image sequence | |
| JP5743935B2 (en) | Object detection apparatus and object detection method | |
| CN110030928A (en) | The method and system of space object positioning and measurement based on computer vision | |
| CN115880643B (en) | Social distance monitoring method and device based on target detection algorithm | |
| JP2019027882A (en) | Object distance detector | |
| Hadi et al. | Fusion of thermal and depth images for occlusion handling for human detection from mobile robot | |
| Yang et al. | Design flow of motion based single camera 3D mapping | |
| CN105631431B (en) | The aircraft region of interest that a kind of visible ray objective contour model is instructed surveys spectral method | |
| KR102028726B1 (en) | Analysis apparatus of object motion in space and control method thereof | |
| JP2013114381A (en) | Object identification device and object identification method | |
| KR101980899B1 (en) | Apparatus for detecting of inside wall frame in single image using orthogonal vanishing points and method thereof | |
| CN106713741B (en) | Panoramic video quality diagnosis method and device | |
| Ringaby et al. | Co-aligning aerial hyperspectral push-broom strips for change detection | |
| Szwoch et al. | Spatial Calibration of a Dual PTZ‐Fixed Camera System for Tracking Moving Objects in Video | |
| WO2018087545A1 (en) | Object location technique |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |