CN1756313A - The panorama composing method of sport video - Google Patents
The panorama composing method of sport video Download PDFInfo
- Publication number
- CN1756313A CN1756313A CN 200410081023 CN200410081023A CN1756313A CN 1756313 A CN1756313 A CN 1756313A CN 200410081023 CN200410081023 CN 200410081023 CN 200410081023 A CN200410081023 A CN 200410081023A CN 1756313 A CN1756313 A CN 1756313A
- Authority
- CN
- China
- Prior art keywords
- video
- frame
- sport
- panorama
- panorama sketch
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 239000000284 extract Substances 0.000 claims abstract description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 238000001914 filtration Methods 0.000 claims description 8
- 230000009471 action Effects 0.000 abstract description 10
- 230000008569 process Effects 0.000 abstract description 5
- 230000008901 benefit Effects 0.000 abstract description 2
- 230000009466 transformation Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 3
- 239000004744 fabric Substances 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 241000272525 Anas platyrhynchos Species 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 230000000386 athletic effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Abstract
The invention discloses a kind of panorama composing method of sport video, be used for a plurality of frame of video of one section sport video are synthesized a panorama sketch, include sport foreground and background in the described sport video, all include sport foreground and part background in each frame of video, include background overall picture and a series of sport foreground in this sport video in the described panorama sketch; This method comprises: the background sprite image of describing background overall picture in the sport video with described a plurality of frame of video generations one; Extract the sport foreground in described a plurality of frame of video respectively; The sport foreground that extracts is added in the background sprite image.The advantage of the method for the invention is: the situation that this method is violent for camera motion or foreground moving is violent is effective too; Prospect is very clear among the result who obtains at last, can well show the process and the details of motion object action.
Description
Technical field
The present invention relates to a kind of image processing method, specifically, relate to a kind of method by the synthetic panorama sketch of sport video.
Background technology
Panorama sketch is meant usually the image of a continuous scenery of description that is generated by a series of images that overlap each other.Panorama sketch is synthetic to be meant by a series of topographies of describing continuous scene synthetic image single, that represent the whole scene overall picture.Utilizing the synthetic panorama sketch of video is that the series of frame images in one section video is synthesized single, as to represent a whole scene overall picture image.The panorama sketch synthetic technology has a wide range of applications in fields such as virtual reality, image encodings.
The panorama sketch synthetic technology can be widely used in sport video.For example, according to the synthetic panorama sketch of the video clips of sports, synkinesia person and coach's competition and training is the synthetic new application of panorama sketch.Sport video includes the prospect of a motion object as video usually, and the image except that prospect in the video is referred to as background.A secondary synthetic back panorama sketch as shown in Figure 1, this panorama sketch has represented a series of actions of the overall picture and the motion object (sportsman among Fig. 1) of background simultaneously in piece image, than allow the sportsman go to watch video shows playback analyze before the gain and loss of action, seem more directly perceived.
Because the particularity of sport video makes the panorama sketch of sport video that the characteristics that tradition is used that are different from be arranged, new requirement proposed to panorama sketch is synthetic.In traditional application, panorama sketch obtains is staticly in the scene not have a mobile background, and in the panorama sketch of sport video, not only want the complete background sprite image that constructs moving scene, more require on background, to show a series of prospect (image of motion object), show the track of motion and the details of action.With respect to video itself, the information that panorama sketch provides is more directly perceived, more comprehensively.Realize that from technology the source images of sport video has that continuity is strong, motion is violent, characteristics such as tangible prospect existence are arranged, and caused the synthetic difficulty of background sprite image bigger.And on the background that generates that how action of the motion object of prospect is added to clearly, also lack deep research at present.
In the panorama composing method of existing sport video, there is the method for the splicing adopted to carry out the synthetic of background sprite image.The splicing method is successively each frame to be projected in the background coordinate system, if certain piece zone assignment is not as yet originally just composed this frame to the background overall situation, this method is just as patch installing more on one point.Very big and video camera moves under the very fast situation in prospect to utilize this method, just cannot say for sure to have some foreground area and is added on the background sprite image by mistake.In addition, the foreground area of first frame can't be eliminated when using the splicing method, also keeps the image of the sport foreground of that frame on complete background sprite image, need eliminate once more with other algorithms, does like this and inconvenient.
Summary of the invention
The objective of the invention is to solve existing panorama sketch synthetic technology and can't overcome that the motion video image continuity is strong, motion is violent, defective such as tangible prospect existence is arranged, a kind of panorama composing method that can show the process and the details of motion object action well is provided.
For achieving the above object, the invention provides a kind of panorama composing method of sport video, be used for a plurality of frame of video of one section sport video are synthesized a panorama sketch, include sport foreground and background in the described sport video, all include sport foreground and part background in each frame of video, include background overall picture and a series of sport foreground in this sport video in the described panorama sketch; This method comprises:
The background sprite image of background overall picture in the sport video is described with described a plurality of frame of video generations one;
Extract the sport foreground in described a plurality of frame of video respectively;
The sport foreground that extracts is added in the background sprite image.
In such scheme, also be included on the frame of video in described a plurality of frame of video and set up a panorama sketch coordinate system, described panorama sketch is synthetic on this panorama sketch coordinate system.
In such scheme, the generation of described background sprite image comprises step:
Global motion relative between described a plurality of frame of video is carried out overall motion estimation;
According to the overall motion estimation between described a plurality of frame of video, described a plurality of frame of video are mapped in the described panorama sketch coordinate system, obtain the scope of described panorama sketch in the panorama sketch coordinate system;
According to the overall motion estimation between described a plurality of frame of video, seek the pixel of each some correspondence in described a plurality of frame of video in the panorama sketch scope; With the pixel value intermediate value of all pixels of searching out, as this pixel value in the panorama sketch, to constitute background sprite image.
In such scheme, described overall motion estimation comprises:
Step adopts the Konrad algorithm that global motion relative between two frame of video is carried out overall motion estimation according to a preliminary estimate, obtains preliminary global motion parameter;
Two step estimating step adopt based on the exterior point filter algorithm of Fisher linear decision rule the preliminary global motion parameter that obtains in the step are according to a preliminary estimate handled, and obtain the global motion parameter.
In such scheme, the described sport foreground of extracting from a plurality of frame of video comprises:
Described a plurality of frame of video are projected to the panorama sketch coordinate system, each frame of video after projection to a projected frame image;
Projected frame image and background sprite image are done the difference computing, obtain a gray level image;
Described gray level image is made binaryzation, obtain a binary image that can be used as prospect-background template;
The described binary image of foundation extracts the sport foreground in the frame of video.
In such scheme, described gray level image is made medium filtering, to eliminate the noise in the gray level image.
In such scheme, described binary image is made morphology and handle, to eliminate the noise in the binary image.
In such scheme, also comprise:
The prospect of a plurality of frame of video of being extracted is superposed to a prospect panorama sketch;
Described prospect panorama sketch is added in the described background sprite image.
In such scheme, when the prospect to a plurality of frame of video superposes, for part overlapping in panorama sketch, adopted the method for ALPHA mixing, and the frame of video after leaning in the frame sequence of forming in described a plurality of frame of video has been not less than the weight of forward frame of video.
Method advantage of the present invention is:
1, the situation that this method is violent for camera motion or foreground moving is violent is effective too.
2, prospect among the result who obtains at last (motion object images) is very clear, can well show the process and the details of motion object action.
Description of drawings
Fig. 1 is an exemplary panorama sketch.
Fig. 2 is the flow chart of the inventive method.
Embodiment
Below in conjunction with the drawings and specific embodiments the present invention is described in further detail.
Fig. 2 shows the flow chart according to the panorama composing method embodiment of a sport video of the present invention.
In the shooting process of sport video, the athletic meeting of video camera causes the motion of video image background, and this forms of motion just is called as global motion.Corresponding with global motion is local motion, and local motion refers to the motion of the prospect that the motion object action causes.Prospect is a pair of relative notion with background, and in the applied sport video of the present invention, prospect is meant the motion object itself in the video, and background is meant removes motion object part in addition in the video.
As shown in Figure 2, in step 101, from one section sport video, extract a plurality of frame of video to be synthesized, the sequence of frames of video that these a plurality of frame of video are normally taken in the video in chronological order.A plurality of frame of video have been described the motion object a series of motion states of sport foreground in background in other words.
In step 102, set up the panorama sketch coordinate system.At first set up the coordinate system of self for each two field picture of a plurality of frame of video, then the coordinate of pixel has reflected the position of this pixel in this two field picture in this two field picture.But because the existence of global motion, the same point in the background has different coordinates in different frame of video.When the structure panorama sketch, all pixels must project to a unified coordinate system, therefore, need set up a unified coordinate system, can be referred to as the panorama sketch coordinate system.Usually, the panorama sketch coordinate system can be chosen to be the coordinate system of arbitrary frame of video in a plurality of frame of video, and this chosen frame of video can be described as reference frame.Preferably, can be with first frame of video in a plurality of frame of video as reference frame, and with the coordinate system of this reference frame as the panorama sketch coordinate system.
In step 103, global motion relative between the frame of video is carried out overall motion estimation, obtain the global motion parameter.Usually, can estimate global motion between every adjacent two frames in a plurality of frame of video, thereby can calculate the global motion of each frame of video by recursion with respect to aforementioned reference frame.Can obtain the transformation parameter of each frame of video coordinate system like this with respect to the panorama sketch coordinate system.
The rule of the global motion between the adjacent image frame can be by the global motion parameter characterization, according to overall motion estimation at the difference of scene, can select different parameter models for use.Select the affine motion model of six parameters in one embodiment of the invention for use:
If (x
i, y
i)
TBe the position of certain pixel in current frame image, (x '
i, y '
i)
TBe the position of this pixel in the previous frame image, then relation between the two can be expressed as
Wherein
The expression video camera convergent-divergent and rotatablely move;
The translational motion of expression video camera.
For A and the T that finds the solution global motion, the present invention adopts the two step estimations technique: the Konrad algorithm that at first utilizes MPEG-4 verification model to be adopted obtains preliminary estimated result; Then,, adopt based on the exterior point filter algorithm of Fisher linear decision rule and carry out the estimation of two steps, obtain overall motion estimation more accurately in order to eliminate the interference of noise datas (being exterior point) such as characteristic matching error, local motion.Compare with the two step estimations technique that the present invention adopts, conventional method is only done estimation with first step algorithm or other similar algorithm, does the influence that is subjected to foreground moving easily like this, and resulting global motion parameter is not too accurate.
Can be about the Konrad algorithm referring to list of references 1:Fr é d é ric Dufaux and JanuszKonrad.Efficient, Robust, and Fast Global Motion Estimation for VideoCoding.IEEE Transactions On Image Processing, Vol.9, No.3, March 2000.
The Konrad algorithm is simply described, and its calculating process is:
1) two two field pictures to input dwindle three pairs of images that form different resolution, form three layers of pyramid structure, and wherein the 3rd layer of resolution is minimum, and second layer resolution is higher, and ground floor is an original image, and its resolution is the highest;
2) the 3rd tomographic image is carried out pre-estimation with the n step search method of revising;
3),, carry out interative computation with the gradient descent method, till an iteration convergence condition is satisfied at pyramidal each layer by from top to bottom order.
After utilizing the Konrad algorithm to carry out according to a preliminary estimate, obtain the preliminary in other words global motion parameter A of result according to a preliminary estimate
*And T
*A according to a preliminary estimate
*And T
*Be that integral image is carried out results estimated, carrying out piece when coupling,, both mixed estimate that resulting result can not effectively describe (overall situation just) motion of background because the motion of the motion of prospect part and background is different.Therefore the present invention also adopts based on the exterior point filter algorithm of Fisher linear decision rule and carries out the estimation of two steps.
The Fisher linear decision rule is a basic conception in the pattern recognition, and it is an algorithm of a series of samples being classified according to feature.Basic skills based on the exterior point filter algorithm of Fisher linear decision rule is:
1) in two field picture, chooses a blocked areas, avoid the image border and most possibly have the image mid portion of prospect, the image block in the blocked areas is carried out the piece match search, obtain a group of feature point { (u
1, u
1') ..., (u
N, u
N').Described match search adopted block matching algorithm general in the general video coding, in the present invention in order to reduce amount of calculation and to improve matching precision, only chooses variance and removes to carry out match search greater than the piece of image variance 1/2nd.Choosing that described characteristic point is right is that the pixel of getting their upper left corner is right as characteristic point in the piece that mates mutually in pairs that finds.
2) obtain an aforesaid group of feature point to { (u
1, u
1') ..., (u
N, u
N') after, the global motion that obtains with the first step in the two step estimations technique is worth A according to a preliminary estimate
*And T
*Calculate the right estimation residual error of each point:
According to this formula, can obtain the right residual error of each point, be feature with the residual error, each can be divided into two set with the Fisher linear decision rule, what residual error was big is exterior point, what residual error was little is interior point.Exterior point is carried out filtering, behind the filtering exterior point, reappraise global motion with some set in the residue.The exterior point that hockets filters and carries out parameter Estimation from remaining interior point, is no more than 10 iteration usually and just can reaches the condition of convergence, obtains estimated result more accurately.
Can see the algorithm of estimating from second step, distinguish prospect and background in estimating in two steps, to calculating global motion, resulting global motion parameter is more accurate from the characteristic point of background.
In step 104, after the global motion parameter between the consecutive frame in a plurality of frame of video that obtain according to step 103, the transformation parameter in the time of then can utilizing the method for recursion iteration to obtain the mutual projection of point between any two frames.Like this, in one embodiment,,, can obtain the alignment transformation parameter A that pixel in first frame projects to the coordinate system of other each frame according to the global motion parameter in the step 103 when with first frame in a plurality of frame of video during as reference frame
I-＞1And T
I-＞1(i=2,3 ... N), utilize alignment transformation parameter A
I-＞1And T
I-＞1, have:
Above this formulate by the coordinate of the point in first frame, utilize corresponding alignment transformation parameter A
I-＞1And T
I-＞1, can be in the hope of the coordinate of this spot projection in the i frame.Equally, also can obtain opposing neat transformation parameter A
1-＞iAnd T
1-＞i(i=2,3 ... N), have
Above this formulate by the coordinate of the point in the i frame, utilize the corresponding neat transformation parameter A of opposition
1-＞iAnd T
1-＞i, can be in the hope of the coordinate of this spot projection in first frame coordinate system.
In step 105, determine panorama sketch to be synthesized scope in the panorama sketch coordinate system.Still be that reference frame is an example with first frame, try to achieve first frame after the alignment of each frame and oppose neat parameter, at first will determine the size of panorama sketch, promptly each frame projects to the scope behind the panorama sketch coordinate system.Specific practice is: get four summits of the 2nd～N frame respectively, try to achieve their coordinates on the panorama sketch coordinate system with the neat conversion of opposition.Get the maximum x of directions X
MaxWith minimum value x
Min, the maximum y of Y direction
MaxWith minimum value y
MinThe wide w=x of panorama sketch then
Max-x
Min, high h=y
Max-y
Min
In step 106, for each point in the panorama sketch all generates an information chained list.By the alignment conversion, can obtain the coordinate of spot projection in each frame coordinate system on the panorama sketch, judge whether to have corresponding points at this frame.For point on the panorama sketch, it both may be the point of a non-image areas, did not promptly all have corresponding points at all frames; Also may only corresponding points be arranged, corresponding points are perhaps all arranged in a plurality of frames at a frame.Each point to panorama sketch all generates a chained list, and this point has corresponding points on which frame, and just the pixel value with these corresponding points adds chained list, and chained list length does not wait (0 expression non-image areas) by 0 to N.For example there are corresponding points in o'clock in the panorama sketch scope in five frame of video, and then the chained list length of this point is exactly five.
In step 107, the generation background panorama sketch.The value that comprises on the chained list of each point in the panorama sketch scope may both be had powerful connections, and prospect and noise are arranged again, just need therefrom try to achieve the value of most possible representative background and compose to panorama sketch.Generally, the number of prospect and noise always is less than the number of background in the chained list, therefore adopts median method to obtain the pixel value of panorama sketch.Median method particularly be exactly with correspondence any all elements the rgb value ordering and its median composed to background sprite image.
Introduce the specific implementation of median method below, suppose on the panorama sketch that certain any chained list is (v
0, v
1, v
2..., v
L) (0＜L＜=N-1), wherein v
i=(R
i, G
i, B
i)
TBe the vector of remarked pixel rgb value, chained list successively sorts according to the frame number of correspondence.The rgb value of elements all in the chained list is sorted, obtain
Get its middle value then and compose, then can represent the information of background substantially to panorama sketch.The number of prospect and noise always is less than the number of background in the chained list owing to mentioned at preamble, is the pixel value of background rather than the pixel value of prospect and noise so get that median can guarantee to compose substantially to background sprite image.
After each point in the panorama sketch scope is all composed with pixel value by said method, then obtain background sprite image.
In step 108, extract the foreground area on a plurality of frame of video.Each frame of video in a plurality of frame of video projects to frame of video on the coordinate system of panorama sketch as affine transformation.Ask four coordinates of summit under the panorama sketch coordinate system of current video frame, the scope after the projection of this frame of acquisition with opposing neat transformation parameter.Within this scope, make difference with projected frame image and background sprite image, obtain the gray level image of an expression difference.Gray-scale map is the image of each pixel monochrome information in the reflection image, and it is used for reflecting the degree that pixel changes between frame and the frame.The formula of difference is:
ε(x，y)＝|R
F(x，y)-R
B(x，y)|+|G
F(x，y)-G
B(x，y)|+|B
F(x，y)-B
B(x，y)|
Wherein x, y are the coordinate of panorama sketch, R
F(x, y), G
F(x, y), B
F(x y) is the rgb value of projected frame image, R
B(x, y), G
B(x, y), B
B(x y) is the background sprite image rgb value, and (x y) is the gray scale of differential chart, ε to D
0Be a threshold value, be used for some Light Differences that filtering causes because of reasons such as illumination, the size of this threshold values is easy to determine for a person skilled in the art.
Have noise in the gray level image that obtains as difference, comprise some little zones and some stripeds.Cause the reason of these noises to mainly contain two: the one, the background sprite image that adopts median method to obtain is equivalent to carry out smothing filtering, than the two field picture of projection fuzzy, can produce bigger difference in the subregion; The 2nd, because the error of global motion parameter makes the two field picture coordinate after the projection also can have error, the edge of object can't align between prospect and the background, produces some stripeds on differential chart.To the closeer zone of image texture, this situation is especially obvious.Therefore, preferred elder generation carries out medium filtering to the gray level image that obtains as difference, removes the isolated noise of those stria shapes and fritter, and the difference to foreground area plays level and smooth effect simultaneously.Then the gray level image behind the medium filtering is done binary conversion treatment, binaryzation is the basic skills in the image processing, finger all becomes brightness in the gray level image greater than the point of certain value white, become less than the point of certain value black, thereby obtain a black and white image.The black and white image that obtains as binaryzation can be understood as the template of a description prospect-background, in this template, and the corresponding background area of black part, the corresponding foreground area of white part.The effect of this template is partly to take out the prospect of corresponding white portion in the two field picture of projection with synthetic prospect panorama sketch.Handle resulting binary image with morphology methods at last: at first, binary image is carried out a closed operation, fill up the duck eye in the foreground area, softening bulk edges of regions; Carry out one time opening operation afterwards again, the noise region of fritter is eliminated.Through morphologic processing, eliminated the noise of fritter substantially, obtained the accurate profile of foreground area, obtain an accurate template in other words.
Last resultant binary image can be extracted the image in the foreground area in each frame of video (being the prospect of frame of video) exactly as template.
In step 109, generate one with the measure-alike base plate of background sprite image, the foreground area that will extract from a plurality of frame of video is added on this base plate, to form a prospect panorama sketch.
After obtaining the foreground area template of each frame of video, also just can obtain the foreground image after the projection.For prevent with background in original pixel interfere with each other, at first generate a prospect panorama sketch that has only each frame prospect to constitute, and then on the background sprite image that is added to.The key of synthetic prospect panorama sketch is exactly how each frame foreground area overlapped part on the prospect panorama sketch is handled.
When a point on the prospect panorama sketch only with to be in the point of foreground area corresponding on the frame to be superimposed, it is overlapping to that is to say that this point does not have, then directly will this frame to be superimposed on the pixel value assignment of corresponding points of foreground area to base plate.On point on the prospect panorama sketch and a plurality of frames to be superimposed, be in a plurality of corresponding of foreground area, that is to say that this point has when overlapping, in the present invention, in order better to embody the course of action of motion object, preferably the pixel value to these a plurality of points carries out ALPHA fusion stack with different weights.Normally, for a plurality of overlapping frames to be superimposed, the just newer frame after leaning on is not less than the weight of forward frame in frame sequence.Such result is that in overlapping part, the action after the motion object leans on seems more clear usually.It is a kind of prior art that ALPHA merges, and its purpose is that the pixel pointwise in two width of cloth images is synthesized piece image, accomplishes the content of original two width of cloth figure of reflection in same width of cloth figure.
In one embodiment,, when prospect is added on the base plate separately with it according to the sequence of frames of video order, give the weight of the pixel value 50% of current stack frame pixel for the overlapping region, and the weight of the pixel value 50% on the base plate that has been added to before giving.。Like this, suppose to have both pixels of words of stack two frames on the same point respectively to account for 1/2; Be superimposed with the 3rd frame then front cross frame respectively account for 1/4, the three frame and account for 1/2; Be superimposed with the 4th frame then front cross frame respectively account for 1/8, the three frame and account for 1/4, the four frame and account for 1/2; And the like.Like this each frame to be superimposed is superposeed up, obtain the prospect panorama sketch.
In step 110, the partial stack except that base plate in the prospect panorama sketch to background sprite image, is obtained final needed panorama sketch.
Claims (9)
1, a kind of panorama composing method of sport video, be used for a plurality of frame of video of one section sport video are synthesized a panorama sketch, include sport foreground and background in the described sport video, all include sport foreground and part background in each frame of video, include background overall picture and a series of sport foreground in this sport video in the described panorama sketch; This method comprises:
The background sprite image of background overall picture in the sport video is described with described a plurality of frame of video generations one;
Extract the sport foreground in described a plurality of frame of video respectively;
The sport foreground that extracts is added in the background sprite image.
2, the panorama composing method of sport video according to claim 1 is characterized in that, also is included on the frame of video in described a plurality of frame of video and sets up a panorama sketch coordinate system, and described panorama sketch is synthetic on this panorama sketch coordinate system.
3, the panorama composing method of sport video according to claim 2 is characterized in that, the generation of described background sprite image comprises step:
Global motion relative between described a plurality of frame of video is carried out overall motion estimation;
According to the overall motion estimation between described a plurality of frame of video, described a plurality of frame of video are mapped in the described panorama sketch coordinate system, obtain the scope of described panorama sketch in the panorama sketch coordinate system;
According to the overall motion estimation between described a plurality of frame of video, seek the pixel of each some correspondence in described a plurality of frame of video in the panorama sketch scope; With the pixel value intermediate value of all pixels of searching out, as this pixel value in the panorama sketch, to constitute background sprite image.
4, the panorama composing method of sport video according to claim 3 is characterized in that, described overall motion estimation comprises:
Step adopts the Konrad algorithm that global motion relative between two frame of video is carried out overall motion estimation according to a preliminary estimate, obtains preliminary global motion parameter;
Two step estimating step adopt based on the exterior point filter algorithm of Fisher linear decision rule the preliminary global motion parameter that obtains in the step are according to a preliminary estimate handled, and obtain the global motion parameter.
5, the panorama composing method of sport video according to claim 2 is characterized in that, the described sport foreground of extracting from a plurality of frame of video comprises:
Described a plurality of frame of video are projected to the panorama sketch coordinate system, each frame of video after projection to a projected frame image;
Projected frame image and background sprite image are done the difference computing, obtain a gray level image;
Described gray level image is made binaryzation, obtain a binary image that can be used as prospect-background template;
The described binary image of foundation extracts the sport foreground in the frame of video.
6, the panorama composing method of sport video according to claim 5 is characterized in that, also comprises described gray level image is made medium filtering, to eliminate the noise in the gray level image.
7, the panorama composing method of sport video according to claim 5 is characterized in that, comprises that also described binary image is made morphology to be handled, to eliminate the noise in the binary image.
8, the panorama composing method of sport video according to claim 1 is characterized in that, also comprises:
The prospect of a plurality of frame of video of being extracted is superposed to a prospect panorama sketch;
Described prospect panorama sketch is added in the described background sprite image.
9, according to the panorama composing method of claim 1 or 8 described sport videos, it is characterized in that, when the prospect to a plurality of frame of video superposes, for part overlapping in panorama sketch, adopted the method for ALPHA mixing, and the frame of video after leaning in the frame sequence of forming in described a plurality of frame of video has been not less than the weight of forward frame of video.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100810236A CN100337473C (en) | 2004-09-30 | 2004-09-30 | Panorama composing method for motion video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100810236A CN100337473C (en) | 2004-09-30 | 2004-09-30 | Panorama composing method for motion video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1756313A true CN1756313A (en) | 2006-04-05 |
| CN100337473C CN100337473C (en) | 2007-09-12 |
Family
ID=36689216
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004100810236A Active CN100337473C (en) | 2004-09-30 | 2004-09-30 | Panorama composing method for motion video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100337473C (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1940152A2 (en) * | 2006-12-27 | 2008-07-02 | Samsung Electronics Co., Ltd. | Method for photographing panoramic image |
| CN101277429B (en) * | 2007-03-27 | 2011-09-07 | 中国科学院自动化研究所 | Method and system for amalgamation process and display of multipath video information when monitoring |
| CN102332158A (en) * | 2011-07-26 | 2012-01-25 | 深圳市万兴软件有限公司 | Image synthesis method and system using alpha channels |
| CN102333221A (en) * | 2011-10-21 | 2012-01-25 | 北京大学 | Panoramic background prediction video coding and decoding method |
| CN102387296A (en) * | 2010-09-03 | 2012-03-21 | 卡西欧计算机株式会社 | Image processing apparatus, method and recording medium for generating a composite image |
| CN102469259A (en) * | 2010-11-01 | 2012-05-23 | 奥林巴斯映像株式会社 | Panoramic image generation device and panoramic image generation method |
| CN103139464A (en) * | 2011-11-29 | 2013-06-05 | 三星电子株式会社 | Method of providing panoramic image and imaging device thereof |
| CN103366351A (en) * | 2012-03-29 | 2013-10-23 | 华晶科技股份有限公司 | Method for generating panoramic image and image acquisition device thereof |
| CN104408701A (en) * | 2014-12-03 | 2015-03-11 | 中国矿业大学 | Large-scale scene video image stitching method |
| CN105096283A (en) * | 2014-04-29 | 2015-11-25 | 华为技术有限公司 | Panoramic image acquisition method and device |
| CN107409231A (en) * | 2015-03-05 | 2017-11-28 | 索尼公司 | Image processing apparatus and image processing method |
| CN107948519A (en) * | 2017-11-30 | 2018-04-20 | 广东欧珀移动通信有限公司 | Image processing method, device and equipment |
| CN108282612A (en) * | 2018-01-12 | 2018-07-13 | 广州市百果园信息技术有限公司 | Method for processing video frequency and computer storage media, terminal |
| CN108921867A (en) * | 2018-06-01 | 2018-11-30 | 天津煋鸟科技有限公司 | A kind of motion detection and human body recognition method of panoramic picture |
| CN109118507A (en) * | 2018-08-27 | 2019-01-01 | 明超 | Shell bears air pressure real-time alarm system |
| CN110782409A (en) * | 2019-10-21 | 2020-02-11 | 太原理工大学 | Method for removing shadow of multi-motion object |
| CN111133763A (en) * | 2017-09-26 | 2020-05-08 | Lg 电子株式会社 | Superposition processing method and device in 360 video system |
| CN111656433A (en) * | 2018-02-03 | 2020-09-11 | 脸谱科技有限责任公司 | Apparatus, system, and method for mitigating motion display delay in head mounted displays |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IL108957A (en) * | 1994-03-14 | 1998-09-24 | Scidel Technologies Ltd | System for implanting an image into a video stream |
| US5515109A (en) * | 1995-04-05 | 1996-05-07 | Ultimatte Corporation | Backing color and luminance nonuniformity compensation |
| US6215505B1 (en) * | 1997-06-20 | 2001-04-10 | Nippon Telegraph And Telephone Corporation | Scheme for interactive video manipulation and display of moving object on background image |
| US6738424B1 (en) * | 1999-12-27 | 2004-05-18 | Objectvideo, Inc. | Scene model generation from video for use in video processing |
| ES2352675T3 (en) * | 2000-04-07 | 2011-02-22 | Dartfish Sa | AUTOMATIC VIDEO-SEQUENCE STROBOSCOPY. |
| US7227896B2 (en) * | 2001-10-04 | 2007-06-05 | Sharp Laboratories Of America, Inc. | Method and apparatus for global motion estimation |
| GB0227565D0 (en) * | 2002-11-26 | 2002-12-31 | British Telecomm | Method and system for generating panoramic images from video sequences |
-
2004
- 2004-09-30 CN CNB2004100810236A patent/CN100337473C/en active Active
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10530979B2 (en) | 2006-12-27 | 2020-01-07 | Samsung Electronics Co., Ltd. | Panoramic digital photographic apparatus including a user interface indicating in real time a suggested direction of movement for the apparatus based on a central portion of a current image |
| US9712734B2 (en) | 2006-12-27 | 2017-07-18 | Samsung Electronics Co., Ltd. | Method for photographing panoramic image without mounting additional components |
| EP1940152A2 (en) * | 2006-12-27 | 2008-07-02 | Samsung Electronics Co., Ltd. | Method for photographing panoramic image |
| EP3062508A1 (en) * | 2006-12-27 | 2016-08-31 | Samsung Electronics Co., Ltd. | Method for photographing panoramic image |
| CN101277429B (en) * | 2007-03-27 | 2011-09-07 | 中国科学院自动化研究所 | Method and system for amalgamation process and display of multipath video information when monitoring |
| CN102387296A (en) * | 2010-09-03 | 2012-03-21 | 卡西欧计算机株式会社 | Image processing apparatus, method and recording medium for generating a composite image |
| CN103856726B (en) * | 2010-09-03 | 2017-04-12 | 卡西欧计算机株式会社 | Image processing device and image processing method |
| CN102469259B (en) * | 2010-11-01 | 2015-06-17 | 奥林巴斯映像株式会社 | Panoramic image generation device and panoramic image generation method |
| CN102469259A (en) * | 2010-11-01 | 2012-05-23 | 奥林巴斯映像株式会社 | Panoramic image generation device and panoramic image generation method |
| CN102332158A (en) * | 2011-07-26 | 2012-01-25 | 深圳市万兴软件有限公司 | Image synthesis method and system using alpha channels |
| CN102332158B (en) * | 2011-07-26 | 2013-06-12 | 深圳万兴信息科技股份有限公司 | Image synthesis method and system using alpha channels |
| CN102333221B (en) * | 2011-10-21 | 2013-09-04 | 北京大学 | Panoramic background prediction video coding and decoding method |
| CN102333221A (en) * | 2011-10-21 | 2012-01-25 | 北京大学 | Panoramic background prediction video coding and decoding method |
| CN103139464A (en) * | 2011-11-29 | 2013-06-05 | 三星电子株式会社 | Method of providing panoramic image and imaging device thereof |
| CN103139464B (en) * | 2011-11-29 | 2018-07-20 | 三星电子株式会社 | The method and its imaging device of panoramic picture are provided |
| CN103366351B (en) * | 2012-03-29 | 2016-06-22 | 华晶科技股份有限公司 | Produce method and the image acquiring device thereof of panoramic picture |
| CN103366351A (en) * | 2012-03-29 | 2013-10-23 | 华晶科技股份有限公司 | Method for generating panoramic image and image acquisition device thereof |
| CN105096283B (en) * | 2014-04-29 | 2017-12-15 | 华为技术有限公司 | The acquisition methods and device of panoramic picture |
| CN105096283A (en) * | 2014-04-29 | 2015-11-25 | 华为技术有限公司 | Panoramic image acquisition method and device |
| CN104408701A (en) * | 2014-12-03 | 2015-03-11 | 中国矿业大学 | Large-scale scene video image stitching method |
| CN107409231A (en) * | 2015-03-05 | 2017-11-28 | 索尼公司 | Image processing apparatus and image processing method |
| CN107409231B (en) * | 2015-03-05 | 2021-05-11 | 索尼公司 | Image processing apparatus, image processing method, and program |
| CN111133763A (en) * | 2017-09-26 | 2020-05-08 | Lg 电子株式会社 | Superposition processing method and device in 360 video system |
| US11575869B2 (en) | 2017-09-26 | 2023-02-07 | Lg Electronics Inc. | Overlay processing method in 360 video system, and device thereof |
| US10997696B2 (en) | 2017-11-30 | 2021-05-04 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Image processing method, apparatus and device |
| CN107948519A (en) * | 2017-11-30 | 2018-04-20 | 广东欧珀移动通信有限公司 | Image processing method, device and equipment |
| CN108282612A (en) * | 2018-01-12 | 2018-07-13 | 广州市百果园信息技术有限公司 | Method for processing video frequency and computer storage media, terminal |
| CN108282612B (en) * | 2018-01-12 | 2021-11-19 | 广州市百果园信息技术有限公司 | Video processing method, computer storage medium and terminal |
| CN111656433A (en) * | 2018-02-03 | 2020-09-11 | 脸谱科技有限责任公司 | Apparatus, system, and method for mitigating motion display delay in head mounted displays |
| CN111656433B (en) * | 2018-02-03 | 2022-09-23 | 元平台技术有限公司 | Apparatus, system, and method for mitigating motion display delay in head mounted displays |
| CN108921867A (en) * | 2018-06-01 | 2018-11-30 | 天津煋鸟科技有限公司 | A kind of motion detection and human body recognition method of panoramic picture |
| CN109118507A (en) * | 2018-08-27 | 2019-01-01 | 明超 | Shell bears air pressure real-time alarm system |
| CN109118507B (en) * | 2018-08-27 | 2019-09-13 | 嵊州市万睿科技有限公司 | Shell bears air pressure real-time alarm system |
| CN110782409A (en) * | 2019-10-21 | 2020-02-11 | 太原理工大学 | Method for removing shadow of multi-motion object |
| CN110782409B (en) * | 2019-10-21 | 2023-05-09 | 太原理工大学 | Method for removing shadow of multiple moving objects |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100337473C (en) | 2007-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100337473C (en) | Panorama composing method for motion video | |
| TWI709107B (en) | Image feature extraction method and saliency prediction method including the same | |
| US11398037B2 (en) | Method and apparatus for performing segmentation of an image | |
| US11455712B2 (en) | Method and apparatus for enhancing stereo vision | |
| US8983178B2 (en) | Apparatus and method for performing segment-based disparity decomposition | |
| JP5985910B2 (en) | Multi-view video processing method and apparatus using hall rendering | |
| CN109671023A (en) | A kind of secondary method for reconstructing of face image super-resolution | |
| CN1460380A (en) | Method for segmenting multi-resolution video objects | |
| JP2009514103A (en) | Split video foreground | |
| CN1694512A (en) | Synthesis method of virtual viewpoint in interactive multi-viewpoint video system | |
| Guo et al. | Dense123'color enhancement dehazing network | |
| CN111145290B (en) | Image colorization method, system and computer readable storage medium | |
| Cai et al. | Guided attention network for object detection and counting on drones | |
| CN110070517A (en) | Blurred picture synthetic method based on degeneration imaging mechanism and generation confrontation mechanism | |
| CN107958441A (en) | Image split-joint method, device, computer equipment and storage medium | |
| CN1150769C (en) | Static image generation method and device | |
| CN113538243B (en) | Super-resolution image reconstruction method based on multi-parallax attention module combination | |
| CN115298708A (en) | Multi-view neural human body rendering | |
| CN113658044A (en) | Method, system, device and storage medium for improving image resolution | |
| Hong et al. | Near-infrared image guided reflection removal | |
| CN1766928A (en) | A kind of motion object center of gravity track extraction method based on the dynamic background sport video | |
| KR20110112143A (en) | A method for transforming 2d video to 3d video by using ldi method | |
| CN105279741A (en) | Image super-resolution reconstruction method and system based on graph-cut algorithm | |
| CN100337472C (en) | Video composing method with motion prospect | |
| CN111383247A (en) | Method for enhancing image tracking stability of pyramid LK optical flow algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| ASS | Succession or assignment of patent right |
Owner name: BEIJING DONGFANGJIANYU INSTITUTE OF CONCRETE SCIEN Owner name: BEIJING XIN'AO CONCRETE GROUP CO., LTD. Free format text: FORMER OWNER: INST. OF COMPUTING TECHNOLOGY, CHINESE ACADEMY OF SCIENCES Effective date: 20110110 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 100080 NO. 6, KEXUEYUAN SOUTH ROAD, ZHONGGUANCUN, HAIDIAN DISTRICT, BEIJING TO: 100020 NO. A-1, SANTAISHAN, XIAOHONGMEN TOWNSHIP, CHAOYANG DISTRICT, BEIJING |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20110110 Address after: 100020 Beijing City three Chaoyang District Xiaohongmen Township Taishan a No. Co-patentee after: Beijing Dongfang Jianyu concrete science and Technology Research Institute Co.,Ltd. Patentee after: BEIJING XINAO CONCRETE GROUP Co.,Ltd. Co-patentee after: BEIJING NEW VOYAGING CO.,LTD. Address before: 100080 Haidian District, Zhongguancun Academy of Sciences, South Road, No. 6, No. Patentee before: Institute of Computing Technology, Chinese Academy of Sciences |
|
| PP01 | Preservation of patent right | ||
| PP01 | Preservation of patent right |
Effective date of registration: 20170721 Granted publication date: 20070912 |
|
| PD01 | Discharge of preservation of patent | ||
| PD01 | Discharge of preservation of patent |
Date of cancellation: 20200721 Granted publication date: 20070912 |
|
| PP01 | Preservation of patent right | ||
| PP01 | Preservation of patent right |
Effective date of registration: 20200721 Granted publication date: 20070912 |
|
| PD01 | Discharge of preservation of patent | ||
| PD01 | Discharge of preservation of patent |
Date of cancellation: 20230721 Granted publication date: 20070912 |
|
| PP01 | Preservation of patent right | ||
| PP01 | Preservation of patent right |
Effective date of registration: 20230721 Granted publication date: 20070912 |