CN106210691A - For the method and apparatus generating anaglyph - Google Patents
For the method and apparatus generating anaglyph Download PDFInfo
- Publication number
- CN106210691A CN106210691A CN201610357234.0A CN201610357234A CN106210691A CN 106210691 A CN106210691 A CN 106210691A CN 201610357234 A CN201610357234 A CN 201610357234A CN 106210691 A CN106210691 A CN 106210691A
- Authority
- CN
- China
- Prior art keywords
- value
- parallax
- concordance
- pixel
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims description 61
- 230000002708 enhancing effect Effects 0.000 claims abstract description 28
- 238000001914 filtration Methods 0.000 claims abstract description 23
- 238000005728 strengthening Methods 0.000 claims abstract description 20
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000008878 coupling Effects 0.000 claims description 8
- 238000010168 coupling process Methods 0.000 claims description 8
- 238000005859 coupling reaction Methods 0.000 claims description 8
- 230000004075 alteration Effects 0.000 claims description 6
- 230000000717 retained effect Effects 0.000 claims description 2
- 230000008569 process Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 8
- 230000002457 bidirectional effect Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- HEFNNWSXXWATRW-UHFFFAOYSA-N Ibuprofen Chemical compound CC(C)CC1=CC=C(C(C)C(O)=O)C=C1 HEFNNWSXXWATRW-UHFFFAOYSA-N 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000004513 sizing Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
- H04N13/128—Adjusting depth or disparity
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
- H04N13/122—Improving the 3D impression of stereoscopic images by modifying image signal contents, e.g. by filtering or adding monoscopic depth cues
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Image Processing (AREA)
- Processing Or Creating Images (AREA)
Abstract
A kind of equipment for generating anaglyph, by the first image of stereo-picture and the second image are performed Stereo matching respectively, generates the first initial parallax image and the second initial parallax image；Retaining filtering by the first initial parallax image and the second initial parallax image being performed respectively edge, generating the first enhancing anaglyph and second and strengthening anaglyph；And then by using the reliability factor of the parallax value of each pixel for the first and second initial parallax images and the reliability factor of parallax value of each pixels for the first and second enhancing anaglyphs, strengthen the parallax value of each pixel for the first and second initial parallax images.
Description
Cross-Reference to Related Applications
This application claims the korean patent application submitted on May 26th, 2015 to Korean Intellectual Property Office
The priority of No. 10-2015-0073252 and rights and interests, incorporated herein by merging entire contents.
Technical field
The present invention relates to the method and apparatus for generating anaglyph, and more particularly, to by more
The method that accurately strengthen the initial parallax obtained by Stereo matching, generates final parallax picture and setting
Standby.
Background technology
Stereo matching is the technology detecting the corresponding point in page at least two images.Stereo matching skill may be used
Art obtains the parallax of the corresponding point between at least two view of image, and calculates figure based on parallax value
The degree of depth of the object in Xiang.Can such as based on image rendering, robot vision and the next generation cognitive
(realization) every field broadcasted utilizes disparity map.
Stereo Matching Technology has been studied much in image processing field, and has studied its various methods.Vertical
Body matching technique style is Global Algorithm scheme and local algorithm scheme.Global Algorithm scheme uses all
Such as confidence spread (belief propagation) and the global optimization method of pattern cut (graph-cut),
And local algorithm scheme is by dividing, based on block, the method that image calculates and optimizes local cost.
Global Algorithm scheme can perform to extract than local algorithm scheme more accurate parallax value relatively, but flower
Taking a lot of time performs to optimize operation, and so almost without being utilized in real time.Recently, based on local
Algorithm arrangement performs initial matching to calculate initial parallax, and is widely used for protecting based on edge
Stay the method that (edge-preserving) filtering strengthens initial parallax.
For edge, there is bidirectional filtering in one of most representative method retaining filtering.Bidirectional filtering
Each pixel of input picture is set as center pixel, and uses there is on the axle of center pixel pre-scale
Present in very little block, scope (range) kernel and territory kernel between neighbor and center pixel are protected
Stay the edge of this image, and other region is performed smooth.That is, bidirectional filtering is while retaining edge
Smooth disparity value, to strengthen due to the obstruction (occlusion) of the marginal portion found when Stereo matching
Or fuzzy (blurring) phenomenon that mismatch occurs.
Edge retain filtering in view of only one of stereo-picture change by Stereo matching calculate initial
Parallax value, and there is a problem in that, when retaining to each independent utility edge of stereo-picture
During filtering, following phenomenon occurs, i.e. need to maintain the most conforming parallax value mismatch each other.
Information above disclosed in this background section is only used for strengthening the understanding of the background of the present invention, and
So it can include being formed without those skilled in the art in the information of prior art known to this country.
Summary of the invention
Made great efforts to make the present invention, with provide have strengthen more accurately by Stereo matching obtain initial
The advantage of parallax, for the method and apparatus that generates anaglyph.
The example embodiment of the present invention provides a kind of by the equipment generation parallax for generating anaglyph
The method of image.The method of described generation anaglyph comprises the steps that by the first image to stereo-picture
Perform Stereo matching respectively with the second image, generate the first initial parallax image and the second initial parallax figure
Picture；Filtering is retained by the first initial parallax image and the second initial parallax image being performed respectively edge,
Generate the first enhancing anaglyph and second and strengthen anaglyph；By using for the first initial parallax
The reliability factor of the parallax value of each pixel of image and each picture for the first enhancing anaglyph
The reliability factor of parallax value of element, strengthens the parallax of each pixel for the first initial parallax image
Value；With by use for the second initial parallax image each pixel parallax value reliability factor and
For the reliability factor of the parallax value of the second each pixel strengthening anaglyph, strengthen for second
The parallax value of each pixel of initial parallax image.
It is right that the described enhancing step for the parallax value of each pixel of the first initial parallax image comprises the steps that
The parallax value of each corresponding point between the first initial parallax image and the second initial parallax image calculates
About first concordance numerical value；For first strengthen that anaglyph and second strengthens between anaglyph every
The parallax value of one corresponding point calculates the about second concordance numerical value；Every for the first initial parallax image
Each pixel that one corresponding point are corresponding, selects corresponding about first concordance numerical value and corresponding about second one
The left and right concordance numerical value among cause property numerical value, there is high reliability；With for the first initial parallax
Each pixel that each corresponding point of image are corresponding, uses the left and right concordance numerical value selected, determines use
Parallax value in each pixel of the first initial parallax image.
It is right that the described determination step for the parallax value of each pixel of the first initial parallax image comprises the steps that
In each pixel of the first initial parallax image, based on selected left and right concordance numerical value calculate for
The left and right concordance reliability values of each pixel；Each pixel of the first initial parallax image is set successively
It is set to center pixel；Calculate and represent have in the block of preliminary dimension on the axle of center pixel each adjacent
The weighted value of the dependency between pixel and center pixel；With the weighted value of the neighbor used in block and
For the left and right concordance reliability values of neighbor, determine the parallax value of center pixel.
The determination step of the parallax value of described center pixel comprises the steps that by the neighbor in combination block
Weighted value and for the left and right concordance reliability values of neighbor, calculates each neighbor
Whole weighted value；One of selection final weighted value of each neighbor in block；With by center pixel
Parallax value change into the parallax value of the neighbor being selected as the final weighted value of one.
The calculation procedure of described weighted value comprises the steps that and uses the weighted value for each neighbor, counts
Calculate the parallax value of each neighbor being used in the first initial parallax image.
The calculation procedure of described weighted value comprises the steps that each phase used in the first initial parallax image
The parallax value of adjacent pixel and the parallax value of each corresponding adjacent pixel in the first enhancing anaglyph,
Calculate the weighted value of each neighbor.
The calculation procedure of described weighted value comprises the steps that the sky used between each neighbor and center pixel
Spacing and aberration, calculate the weighted value of each neighbor.
The calculation procedure of described weighted value comprises the steps that and uses the cost used for sectional perspective coupling to tire out
Add function, calculate the weighted value of each neighbor.
The described calculation procedure for the left and right concordance reliability values of each pixel comprises the steps that when selecting
Left and right concordance numerical value when being 0, left and right concordance reliability values is calculated as 1；When the left side selected
When right uniformity numerical value is more than threshold value, left and right concordance reliability values is calculated as 0；With when select
When left and right concordance numerical value is equal to or less than threshold value, left and right concordance reliability values is calculated as " 1/ choosing
The left and right concordance numerical value selected " value.
The calculation procedure of described about first concordance numerical value comprise the steps that according to the first initial parallax image and
The difference of the absolute value of the parallax value for each corresponding point between the second initial parallax image, calculates
About one concordance numerical value, and the calculation procedure of described about second concordance numerical value comprises the steps that according to
One strengthens the exhausted of the parallax value for each corresponding point between anaglyph and the second enhancing anaglyph
Difference to value, calculates the about second concordance numerical value.
Described selection step can include the about first concordance numerical value and corresponding about second concordance number
Smaller value in value is defined as the left and right concordance numerical value with high reliability.
One of described first image and the second image can be left images, and another is right image.
Another embodiment of the present invention provides a kind of equipment for being generated anaglyph by stereo-picture.
The described equipment for generating anaglyph can include Stereo matching unit, edge retain filter unit and
Parallax enhancement unit.This Stereo matching unit can be by dividing the first image of stereo-picture and the second image
Do not perform Stereo matching, generate the first initial parallax image and the second initial parallax image.This edge is protected
Stay the filter unit can be by the first initial parallax image and the second initial parallax image are performed edge respectively
Retain filtering, generate the first enhancing anaglyph and second and strengthen anaglyph.This parallax enhancement unit
Can be to two pixels of each corresponding point between the first initial parallax image and the second initial parallax image
Parallax value perform left and right consistency check, to first strengthen anaglyph and second strengthen anaglyph it
Between each corresponding point two pixels parallax value perform left and right consistency check, calculate represent to increase
Have on the axle of strong pixel between each neighbor and the pixel to be strengthened in the block of preliminary dimension
The weighted value for each neighbor of dependency, and use respectively the consistency check of described left and right with
The weighted value of each neighbor in the first initial parallax image and the second initial parallax image, strengthens
The parallax value of each pixel.
This parallax enhancement unit can include left and right consistency checker and parallax value determiner.About this unanimously
Property detector can calculate initially regard for the first initial parallax image and second based on left and right consistency check
About first concordance numerical value of the parallax value of each corresponding point between difference image and for the first increasing
Strong anaglyph and second strengthens about the second one of the parallax value of each corresponding point between anaglyph
Cause property numerical value.This parallax value determiner can be for corresponding with each corresponding point of the first initial parallax image
Each pixel, uses among corresponding about first concordance numerical value and corresponding about second concordance numerical value
, the left and right concordance numerical value with higher reliability and the weighted value of each corresponding adjacent pixel,
Determine the parallax value of each pixel for the first initial parallax image.
This parallax enhancement unit can farther include: left and right concordance reliability calculator, for for often
One pixel, uses left and right concordance numerical value to calculate left and right concordance reliability values；Calculate with weighted value
Device, for for each pixel, by combining weighted value and the left and right concordance of each corresponding adjacent pixel
Reliability values, calculates the final weighted value of each neighbor in corresponding blocks, and this parallax value is true
Determine the final weighted value that device for each pixel, can determine each neighbor having in corresponding blocks
The parallax value of the pixel of the final weighted value selected is as the parallax value for each pixel.
This left and right concordance reliability calculator, when the left and right concordance numerical value with high reliability is 0,
Left and right concordance reliability values can be calculated as 1, when the left and right concordance numerical value with high reliability is big
When threshold value, left and right concordance reliability values can be calculated as 0, and when having the left and right of high reliability
Left and right concordance reliability values, when concordance numerical value is equal to or less than threshold value, can be calculated as " 1/ selection
Left and right concordance numerical value " value.
This parallax enhancement unit can farther include: left and right concordance value selector, for for the
Each pixel that each corresponding point of one initial parallax image are corresponding, determines corresponding about first concordance number
The smaller value of value and corresponding about second concordance numerical value is as the left and right concordance number with high reliability
Value, and this left and right consistency checker is according to the difference of the absolute value of the parallax value for each corresponding point, comes
Calculate the about first concordance numerical value and the about second concordance numerical value.
This parallax enhancement unit can calculate the parallax value of each neighbor as the first initial parallax
The weighted value of each neighbor in image, or use in the first initial parallax image each adjacent
The parallax value of pixel and first strengthens the parallax value of each corresponding adjacent pixel in anaglyph, calculates
The weighted value of each neighbor.
This parallax enhancement unit can use between each neighbor and pixel to be strengthened and center pixel
Space length and aberration calculate the weighted value of each neighbor, or use for sectional perspective
Join used cost sum function to calculate the weighted value of each neighbor.
This parallax value determiner can be for corresponding with each corresponding point of the second initial parallax image each
Pixel, uses among corresponding about first concordance numerical value and corresponding about second concordance numerical value, tool
There are the left and right concordance numerical value of higher reliability and the weighted value for each corresponding adjacent pixel, come
Determine the parallax value of each pixel for the second initial parallax image.
Accompanying drawing explanation
Fig. 1 be schematically illustrate according to example embodiment of the present invention for the method generating anaglyph
Figure.
Fig. 2 depicts the flow chart of the process strengthening anaglyph according to example embodiment of the present invention.
Fig. 3 depicts the conforming figure in the left and right according to example embodiment of the present invention.
Fig. 4 be a diagram that have preliminary dimension on the axle of center pixel according to example embodiment of the present invention
The figure of block.
Fig. 5 be a diagram that the figure of the equipment for generating anaglyph according to example embodiment of the present invention.
Fig. 6 be a diagram that the figure of the parallax enhancement unit of diagram in Fig. 5.
Detailed description of the invention
In the following detailed description, being only used as explanation, some the most only illustrate and describing the present invention is shown
Model embodiment.As the skilled person will recognize, described by can revising in various different ways
Embodiment, and all without departing from the spirit or scope of the present invention.Therefore, these figures and description should be seen
Make the most schematic rather than restrictive.Running through this description, identical reference specifies phase
Same element.
Run through specification and claims, be expressly recited except in contrary mode, " including "
Any assembly will be understood as inferring and includes other element, and is not excluded for other element any.
Thereafter, by be described in detail with reference to the accompanying drawings according to example embodiment of the present invention be used for generate disparity map
The method and apparatus of picture.
Fig. 1 be schematically illustrate according to example embodiment of the present invention for the method generating anaglyph
Figure.
With reference to Fig. 1, for generating the equipment 100 of anaglyph left image and the right side respectively to stereo-picture
Image performs Stereo matching, to generate initial left view difference image and initial right anaglyph (S110).Three-dimensional
Coupling is that one of the left image of stereo-picture and right image are set to reference picture, by left image and right figure
Another of picture is set to target image and then finds out corresponding to each other in reference picture and target image
The operation of corresponding point, and obtain that apart how far represent the respective pixel in left image and right image
Parallax value.Such as, if passing through the Stereo matching for left image, position (x, y) parallax of the pixel at place
Value is 10, then based on the position in right image, (x, y) on the right at a distance of (x+10, picture y) of 10 pixels
Element becomes position (x, the corresponding point of pixel y) of left image.
For generating the equipment 100 of anaglyph to initial left view difference image and initial right anaglyph difference
Perform such as edge and retain the post processing and filtering of filtering, strengthen left view difference image to generate and strengthen right parallax
Image (S120).Such as, retain filtering as edge, bidirectional filtering can be used.Bidirectional filtering can pass through
Equation 1 below represents.
(equation 1)
In above equation 1, DiRepresent the initial left view difference image as input picture or the initial right side regards
Difference image, and the W representation weight values factor.DoRepresent the enhancing left view difference image as output or increasing
Strong right anaglyph.(x y) represents center pixel.Each pixel of input picture is arranged to center pixel.
U and v is the vectorial U representing the ad-hoc location in filter kernelPAnd VPElement.Vector UPAnd VP
Can be represented by equation 2 below and 3.
(equation 2)
(equation 3)
Here, filter kernel represents the block on the axle of center pixel with preliminary dimension, and r represents
The radius of filter kernel.Vector UPAnd VPEach represent the horizontal and vertical in filter kernel
Position.
Additionally, weighted value factor W can be defined as equation 4 below.
(equation 4)
In above equation 4, f (u, v) represents and depends on the territory kernel that space length determines, and g (u,
V) scope that the aberration between the center pixel in filter kernel and neighbor determines is depended in representative
Wave filter.Both the two functions can be defined as equation 5 below and 6 based on Gaussian function.
(equation 5)
(equation 6)
In above equation 5 and 6, σRAnd σDEach represent for the standard deviation of Gaussian function,
And depend on that these values are to determine the width of Gauss distribution.
So, if respectively initial left view difference image and initial right anaglyph being performed edge to retain filter
Ripple, then retain the boundary member that wherein aberration is big, and smooth remainder.
The process initial left view difference image of enhancing is strengthened based on parallax for generating the equipment 100 of anaglyph
(S130).Parallax strengthens process and uses initial left view difference image, initial right anaglyph, enhancing left view poor
Image and strengthen right anaglyph, and will describe in detail with reference to Fig. 2 and 3.
The process initial right anaglyph of enhancing is strengthened based on parallax for generating the equipment 100 of anaglyph
(S140).For the ease of describing, although the example embodiment of the present invention describes the initial left disparity map of enhancing
Then picture also strengthens initial right anaglyph, but can strengthen initial right anaglyph and the most also can strengthen
Initial left view difference image.
Can be by preset frequency N respectively to initial left view difference image for generating the equipment 100 of anaglyph
It is repeated in parallax with initial right anaglyph and strengthens process.
That is, if using parallax to strengthen process strengthen left view difference image and right anaglyph, then it is used for generating
Parallax is strengthened frequency i and increases by 1 (S150) and then compare parallax enhancing frequency by the equipment 100 of anaglyph
Rate i and preset frequency N (S160).It follows that for generating the equipment 100 of anaglyph respectively to just
Beginning left view difference image and initial right anaglyph are repeated in parallax and strengthen process, until parallax strengthens frequency
Till i reaches preset frequency N, to generate final left view difference image and final right anaglyph (S170).
Fig. 2 depicts the flow chart of the process strengthening anaglyph according to example embodiment of the present invention,
And Fig. 3 depicts the conforming figure in the left and right according to example embodiment of the present invention.
First, it is assumed that given initial left view difference image and initial right anaglyph and strengthen left view difference image and
With the right anaglyph of enhancing.Initial left view difference image and initial right anaglyph are collectively referred to as initial parallax figure
Picture, and strengthen left view difference image and be collectively referred to as strengthening anaglyph with strengthening right anaglyph.Additionally,
For the ease of explaining, the method being used for description strengthening left view difference image.
With reference to Fig. 2, for generating the equipment 100 of anaglyph based on initial left view difference image and the initial right side
Anaglyph, checks left and right concordance (S200) to the parallax value of initial parallax image.Left and right concordance inspection
Look into and can perform by calculating left and right concordance numerical value.
Based on enhancing left view difference image and right anaglyph is strengthened for generating the equipment 100 of anaglyph,
The parallax value strengthening anaglyph is checked left and right concordance (S210).Left and right consistency check can be by meter
Calculate left and right concordance numerical value to perform.
As it has been described above, parallax value represents corresponding point and the x in right image of a point of left image by value
Axle is the most how many.Due in addition to blocked-off region, each pixel of left image and right image the most one by one
Join, so each pixel of the left image corresponding with corresponding point and right image needs to point to each other.So,
The absolute value of the parallax value of one point of initial left view difference image is equal to the correspondence in initial right anaglyph
The absolute value of parallax value of point, and the absolute value of parallax value of a point about initial left view difference image
Whether it is referred to as left and right equal to the inspection of the absolute value of the parallax value of the corresponding point in initial right anaglyph
Consistency check.
I.e., as illustrated in Figure 3, if initial left view difference image DLThe parallax value of a some x be
DLX (), then with initial right anaglyph DRA some x at a distance of DLX the point of () pixel becomes for just
Beginning left view difference image DLA some x, initial right anaglyph DRIn corresponding point.Additionally, such as
Fruit and initial right anaglyph DRA point (x) at a distance of DLPoint [the x+D of (x) pixelL(x)] parallax value
For DR(x '), then work as DR(x ')=DLTime (x), it is believed that guarantee left and right concordance.
Can be based on initial left view difference image by equation 7 below root for generating the equipment 100 of anaglyph
The difference of absolute value according to the parallax value of the corresponding point between initial left view difference image and initial right anaglyph
Calculate left and right concordance numerical value, to perform left and right consistency check.
(equation 7)
δ=| DL(x)|-|DR(x+DL(x))|
In above equation 7, DLX () represents the parallax of the pixel at the position x in initial left view difference image
Value, and DR(x+DL(x)) represent the pixel at position [x+DL (x)] place in initial right anaglyph
Parallax value.
By same procedure, can lead to based on strengthening left view difference image for generating the equipment 100 of anaglyph
Cross equation 8 below according to strengthening left view difference image and the parallax value of corresponding point strengthened between right anaglyph
The difference of absolute value calculate left and right concordance numerical value, to perform left and right consistency check.
(equation 8)
δ '=D 'L(x)|-|D′R(x+D′L(x))|
In above equation 8, DL' (x) represent regarding of the pixel at the position x strengthened in left view difference image
Difference, and DR’(x+DL(x)) represent the picture at position [x+DL (the x)] place strengthened in right anaglyph
The parallax value of element.
Owing to the initial parallax image of such as initial left view difference image and initial right anaglyph is based on conjunction
Reason assumes the prediction parallax value that calculated by matching treatment, even if so the parallax value of initial parallax image not
Being same as standing the parallax value of the post processing and filtering that such as edge retains filtering, this value can not also be counted as nothing
Condition errors.If initial parallax image being performed edge retain filtering, then wherein left and right can be occurred consistent
The situation of property mismatch.So, the parallax value of initial parallax image occurs and strengthens the parallax value of anaglyph
In which more reliable problem.
For generating the equipment 100 of anaglyph by every for initial left view difference image of equation 9 below
One pixel, selects having relatively between the parallax value of initial parallax image and the parallax value strengthening anaglyph
Left and right concordance numerical value (S220) of high reliability.Can be based on the left and right calculated by above equation 7 and 8
Concordance numerical value determines high reliability, and along with consistent with the left and right that 8 calculate by above equation 7
Property numerical value just becoming more and more less, reliability increase.
(equation 9)
Δ=argmin (δ, δ ')=(δ ＞ δ '？δ ': δ)
Use, for generating the equipment 100 of anaglyph, the parallax value selected by above equation 9, count
Calculate left and right concordance reliability values (S230) for each pixel.Left and right concordance reliability values
CLRX () can be calculated by equation 10 below.
(equation 10)
With reference to above equation 10, when left and right concordance value, Δ is more than predetermined threshold T, left and right is consistent
Property reliability values CLRX () becomes 0, and when Δ is 0, it is achieved left and right concordance, and tie
Really concordance reliability values CLRX () becomes 1.Additionally, when Δ is equal to or less than predetermined threshold T,
Concordance reliability values CLRX () becomes 1/ Δ being inversely proportional to left and right concordance numerical value.
If calculating left and right concordance reliability values C for each pixel by equation 10LR(x),
Object pixel to be strengthened for its parallax value is set to central authorities' picture by the equipment 100 being then used for generating anaglyph
Element (S240), and calculate left and right to be utilized concordance reliability values, have on the axle of center pixel pre-
The weighted value (S250) that all pixels in the block of sizing add.For generating the equipment of anaglyph
100 can use the cost sum function of all pixels in block to calculate weighted value, with by reference
On the pixel value of center pixel or the axle of parallax value, there is the neighbor in the block of preliminary dimension, calculate
Its dependency or its similarity.Can use from cost add up candidate set select cost sum function calculate
This weighted value.
Cost adds up the cost sum function of candidate set such as equation 11 below to 16.
(equation 11)
W (x)=| DL(x)|or|D′L(x)|
(equation 12)
W (x)=| DL(x)-D′L(x)|
(equation 13)
(equation 14)
W (x)=g (x) f (x)
Here, g (x) and f (x) is as in above equation 5 and 6.
(equation 15)
One of the cost sum function of w (x)=use in sectional perspective is mated
(equation 16)
W (x)=from each of the cost sum function of the combination of at least two including equation 11 to 15
The weighted value obtained is long-pending
For generating what the equipment 100 of anaglyph calculated for each pixel to the weighted value addition calculated
Left and right concordance reliability values, with each pixel in calculating block on the axle of center pixel
Final weighted value (S260).This final weighted value can be calculated by equation 17 below.
(equation 17)
W (q)=w (q) × CLR(q)
In above equation 17, W (q) represents the neighbor q in the block on the axle of center pixel
Whole pixel value, w (q) represents the weight obtained for neighbor q based on one of above equation 11 to 16
Value, and CLRQ left and right concordance that () represents neighbor q based on above equation 10 acquisition is reliable
Property.
If the final weighted value of each pixel in the block calculated on the axle of center pixel, then it is used for generating
The equipment 100 of anaglyph is according to each pixel in the block on the axle of size order classification center pixel
Final weighted value, and select most suitable final weighted value (S270) from the final weighted value of classification.
Such as, for generate anaglyph equipment 100 can also selection sort be the most in some cases
The intermediate value of the weighted value among whole weighted value, and also can weigh by the final of selection sort in some cases
Maximum among weight values or minima.
The pixel of most suitable final weighted value will be selected as generating the equipment 100 of anaglyph
Parallax value is defined as the parallax value (S280) of center pixel, and the parallax value of center pixel is changed into selection
Parallax value (S290) for the pixel of most suitable final weighted value.Can be represented by equation 18 below
For the method changing the parallax value of center pixel.
(equation 18)
D (p)=d of f (W (q1), W (q2) ..., W (qn))
In above equation 18, d represents parallax value and d (p) represents the parallax value of center pixel (p).
W(q1)、W(q2)……W(qn) represent the final weight of each pixel in the block on the axle of center pixel
Value.Additionally, f () represents the most suitable final weighted value among the final weighted value of each pixel in block.
Such as, as it has been described above, f () can be to select the intermediate value among the final weighted value of each pixel in block
Function, and can be to select the maximum among the final weighted value of each pixel in block or minima
Function.
So, in order to be obtained the parallax value for center pixel by the equipment 100 being used for generating anaglyph,
Left and right concordance in view of each pixel in the block with preliminary dimension on the axle of center pixel can
Calculate the final weighted value of each pixel by property numerical value, and will be selected to be each pixel in block
The parallax value of the pixel of the most suitable final weighted value among whole weighted value is defined as regarding of center pixel
Difference.
For generating the equipment 100 of anaglyph, all pixels of initial left view difference image are respectively set to
Center pixel, and the parallax value of center pixel is then determined by performing step S250 to S280, by
This strengthens the parallax value of all pixels for initial left view difference image.
Strengthening in initial right anaglyph performs also by the method described with reference to Fig. 2, and with reference to just
Begin right anaglyph.
Additionally, strengthen the parallax value of initial left view difference image for generating the equipment 100 of anaglyph, and
The parallax value being then based on initially strengthening left view difference image repeats step S230 to S290, thus adds and holds
Row parallax strengthens process.That is, in initial situation, due to given initial left view difference image and enhancing left view
Difference image, so performing step S210 and S220 to select initial left view difference image and to strengthen left disparity map
One of as, and after performing a parallax enhancing process, the left view that generation has enhancing parallax value is poor
Image, and so parallax enhancing process thereafter not necessarily performs step S210 and S220.
Fig. 4 be a diagram that have preliminary dimension on the axle of center pixel according to example embodiment of the present invention
The figure of block.
With reference to Fig. 4, strengthen initial left view difference image D when existingLParallax value D of pixel pLThe need of (p)
When asking, pixel p is set in center pixel p, and the axle of center pixel p being provided as center
The block with preliminary dimension.This block can be such as rectangle and foursquare rigid structure and asymmetric or
Deformable structure, and it is not particularly limited the form of block.
Additionally, even for strengthening left view difference image DLThe pixel p at same position in ', is arranged
As the block with preliminary dimension on the axle of center pixel p at center.
As reference Fig. 2 is for initial left view difference image DLWith enhancing left view difference image DL' each
Each piece of interior pixel q1、q2、……、qnDescribe, weight can be calculated by use cost sum function
Value.
Fig. 5 be a diagram that the figure of the equipment for generating anaglyph according to example embodiment of the present invention,
And Fig. 6 be a diagram that the figure of the parallax enhancement unit of diagram in Fig. 5.
With reference to Fig. 5, include Stereo matching unit 110, edge for generating the equipment 100 of anaglyph
Retain filter unit 120 and parallax enhancement unit 130.
Stereo matching unit 110 performs three-dimensional to each of the left image of stereo-picture and right image
Join, to generate initial left view difference image and initial right anaglyph.As Stereo matching, there is various side
Case, the most block-based coupling, coupling based on pixel, feature-based matching etc..Such as, based on
The coupling of pixel be find out in the target image corresponding with each of all pixels in reference picture
The operation of corresponding point.Block-based coupling is to divide image and at target figure with the block with fixed dimension
The operation of the corresponding point corresponding with each piece is found out in Xiang.
Edge retains the filter unit 120 initial left view difference image to three-dimensional matching unit 110 generation with just
Begin right anaglyph perform respectively edge retain filtering, with generate strengthen left view difference image and strengthen right parallax
Image.Retain filters solutions as edge, the various schemes including bidirectional filtering can be used.
Parallax enhancement unit 130 performs use respectively to initial left view difference image and initial right anaglyph
Initial left view difference image and initial right anaglyph and enhancing left view difference image and strengthen right anaglyph
Parallax strengthen process.Parallax enhancement unit 130 includes that left and right consistency checker 131, left and right are consistent
Property value selector 132, left and right concordance reliability calculator 133, weighted value computer 134, weight
Value selector 135 and parallax value determiner 136.
Left and right consistency checker 131 calculate as above equation 7 from initial left view difference image and
The left and right concordance numerical value of the parallax value of two pixels of the corresponding point of initial right anaglyph.Additionally, it is left
Right uniformity detector 131 calculates self-reinforcing left view difference image as above equation 8 and strengthens the right side
The left and right concordance numerical value of the parallax value of two pixels of the corresponding point of anaglyph.
Left and right concordance value selector 132 selects at the beginning of each pixel as above equation 9
The left and right with high reliability between parallax value and the parallax value strengthening anaglyph of beginning anaglyph
Concordance numerical value.
Left and right concordance reliability calculator 133 uses the left and right concordance selected by above equation 9
Numerical value, calculates the left and right concordance reliability values for each pixel by above equation 10.
Weighted value computer 134 calculates will be to the left and right concordance reliability values calculated for each pixel
The weighted value for each pixel added.Weighted value computer 134 can calculate dependency or similarity,
Can be by adjacent with reference to have in the block of preliminary dimension on the axle of the pixel value of center pixel or parallax value
Pixel, uses the cost sum function of all pixels being used in block, calculates weighted value.
Weighted value computer 134 adds each of which left side calculated for each pixel to the weighted value calculated
Right uniformity reliability values, to calculate in block based on center pixel as above equation 17
The final weighted value of each pixel.
Weighted value selector 135 is according to each pixel in the block on the size order axle to center pixel
Final weighted value classify, and the most suitable final weight among the final weighted value of selection sort
Value.
The parallax value being selected as the pixel of most suitable final weighted value is determined by parallax value determiner 136
For the parallax value of center pixel, and the parallax value of center pixel is changed into it is chosen as most suitable weighted value
The parallax value of pixel, as above equation 18.
The method and apparatus for generating anaglyph according to example embodiment of the present invention described above
At least some function can be realized by hardware or the software coupled with hardware.Such as, it is achieved process for central authorities
Unit (CPU), the processor of other chipset, microprocessors etc. can perform Stereo matching unit 110,
Edge retains filter unit 120 and the function of parallax enhancement unit 130.
According to the example embodiment of the present invention, the parallax calculated by Stereo matching may be given wherein
Image and retain under the state of anaglyph that smothing filtering smooth by edge, by use the two
Anaglyph supplements left and right concordance, and generates anaglyph more accurately.
The example embodiment of the present invention realizes not only by the said equipment and/or method, and can be by realizing
The program of the function corresponding with the configuration of example embodiment of the present invention or record have the record medium of this program
Realizing, it can be easy according to the description of previous exemplary embodiment by those skilled in the art in the invention
Ground realizes.
Currently it is counted as the content of practical demonstration embodiment describes the present invention although having combined, but should manage
Solve, the invention is not restricted to disclosed embodiment, but on the contrary, it is intended to cover claims
The various amendments included in spirit and scope and equivalent arrangement.
Claims (20)
1. by a method for the equipment generation anaglyph for generating anaglyph, including:
By the first image of stereo-picture and the second image are performed Stereo matching respectively, generate first
Initial parallax image and the second initial parallax image；
Filtering is retained by the first initial parallax image and the second initial parallax image being performed respectively edge,
Generate the first enhancing anaglyph and second and strengthen anaglyph；
By use for the first initial parallax image each pixel parallax value reliability factor and
For the reliability factor of the parallax value of the first each pixel strengthening anaglyph, strengthen for first
The parallax value of each pixel of initial parallax image；With
By use for the second initial parallax image each pixel parallax value reliability factor and
For the reliability factor of the parallax value of the second each pixel strengthening anaglyph, strengthen for second
The parallax value of each pixel of initial parallax image.
Method the most according to claim 1, wherein:
The described enhancing step for the parallax value of each pixel of the first initial parallax image includes:
Parallax for each corresponding point between the first initial parallax image and the second initial parallax image
Value calculates the about first concordance numerical value；
Strengthen anaglyph and second for first and strengthen the parallax of each corresponding point between anaglyph
Value calculates the about second concordance numerical value；
For each pixel corresponding with each corresponding point of the first initial parallax image, select corresponding first
The left and right among left and right concordance numerical value and corresponding about second concordance numerical value, there is higher reliability
Concordance numerical value；With
For each pixel corresponding with each corresponding point of the first initial parallax image, use the left side selected
Right uniformity numerical value, determines the parallax value of each pixel for the first initial parallax image.
Method the most according to claim 2, wherein:
The described determination step for the parallax value of each pixel of the first initial parallax image includes:
For each pixel of the first initial parallax image, the axle of the left and right concordance numerical value selected by calculating
On the left and right concordance reliability values for each pixel；
Each pixel of the first initial parallax image is set gradually as center pixel；
Calculate represent have on the axle of center pixel each neighbor in the block of preliminary dimension with in
The weighted value of the dependency between the pixel of centre；With
Use the weighted value of the neighbor in block and be used for the left and right concordance reliability number of neighbor
Value, determines the parallax value of center pixel.
Method the most according to claim 3, wherein:
The determination step of the parallax value of described center pixel includes:
Weighted value by the neighbor in combination block and the left and right concordance for neighbor are reliable
Property numerical value, calculates the final weighted value of each neighbor；
One of selection final weighted value of each neighbor in block；With
The parallax value of center pixel is changed into the neighbor being selected as the final weighted value of one
Parallax value.
Method the most according to claim 3, wherein:
The calculation procedure of described weighted value includes: uses the weighted value for each neighbor, calculates
The parallax value of each neighbor in the first initial parallax image.
Method the most according to claim 3, wherein:
The calculation procedure of described weighted value includes: it is each adjacent that use is used in the first initial parallax image
The parallax value of pixel and the parallax value of each corresponding adjacent pixel in the first enhancing anaglyph, come
Calculate the weighted value of each neighbor.
Method the most according to claim 3, wherein:
The calculation procedure of described weighted value includes: use the space between each neighbor and center pixel
Distance and aberration, calculate the weighted value of each neighbor.
Method the most according to claim 3, wherein:
The calculation procedure of described weighted value includes: use the cost sum function for sectional perspective coupling,
Calculate the weighted value of each neighbor.
Method the most according to claim 3, wherein:
The described calculation procedure for the left and right concordance reliability values of each pixel includes:
When the left and right concordance numerical value selected is 0, left and right concordance reliability values is calculated as 1；
When the left and right concordance numerical value selected is more than threshold value, left and right concordance reliability values is calculated as
0；With
When the left and right concordance numerical value selected is equal to or less than threshold value, by left and right concordance reliability values
It is calculated as the value of " the 1/ left and right concordance numerical value selected ".
Method the most according to claim 2, wherein:
The calculation procedure of described about first concordance numerical value includes: according to the first initial parallax image and
The difference of the absolute value of the parallax value for each corresponding point between two initial parallax images, calculates first
Left and right concordance numerical value, and
The calculation procedure of described about second concordance numerical value includes: strengthen anaglyph and the according to first
The difference of two absolute values strengthening the parallax value for each corresponding point between anaglyph, calculates second
Left and right concordance numerical value.
11. methods according to claim 10, wherein:
Described selection step includes the about first concordance numerical value and corresponding about second concordance numerical value
In smaller value be defined as the left and right concordance numerical value with high reliability.
12. methods according to claim 1, wherein:
One of described first image and the second image are left images, and another is right image.
13. 1 kinds of equipment being used for being generated anaglyph by stereo-picture, including:
Stereo matching unit, by performing three-dimensional respectively to the first image of stereo-picture and the second image
Join, generate the first initial parallax image and the second initial parallax image；
Edge retains filter unit, by the first initial parallax image and the second initial parallax image difference
Perform edge and retain filtering, generate the first enhancing anaglyph and second and strengthen anaglyph；And
Parallax enhancement unit, to every a pair between the first initial parallax image and the second initial parallax image
The parallax value of two pixels that should put performs left and right consistency check, strengthens anaglyph and second to first
The parallax value of two pixels strengthening each corresponding point between anaglyph performs left and right consistency check,
Calculate and represent each neighbor having in the block of preliminary dimension on the axle of pixel to be strengthened and want
The weighted value for each neighbor of the dependency between the pixel strengthened, and use a described left side respectively
Right uniformity checks and each neighbor in the first initial parallax image and the second initial parallax image
Weighted value, strengthen the parallax value of each pixel.
14. equipment according to claim 13, wherein:
This parallax enhancement unit includes:
Left and right consistency checker, for based on left and right consistency check, calculates for the first initial parallax
About first concordance number of the parallax value of each corresponding point between image and the second initial parallax image
Value and regarding for each corresponding point between the first enhancing anaglyph and the second enhancing anaglyph
About second concordance numerical value of difference；With
Parallax value determiner, for for corresponding with each corresponding point of the first initial parallax image each
Pixel, uses among corresponding about first concordance numerical value and corresponding about second concordance numerical value, tool
There are left and right concordance numerical value and the weighted value of each corresponding adjacent pixel of higher reliability, determine
Parallax value for each pixel of the first initial parallax image.
15. equipment according to claim 14, wherein:
This parallax enhancement unit farther includes:
Left and right concordance reliability calculator, for for each pixel, uses left and right concordance numerical value
Calculate left and right concordance reliability values；With
Weighted value computer, for for each pixel, by combining the weight of each corresponding adjacent pixel
Value and left and right concordance reliability values, calculate the final weight of each neighbor in corresponding blocks
Value, and
This parallax value determiner for each pixel, determines each neighbor of having in corresponding blocks
The parallax value of the pixel of the final weighted value selected in final weighted value is as each pixel
Parallax value.
16. equipment according to claim 15, wherein:
This left and right concordance reliability calculator, when the left and right concordance numerical value with high reliability is 0,
Left and right concordance reliability values is calculated as 1, when the left and right concordance numerical value with high reliability is more than
During threshold value, left and right concordance reliability values is calculated as 0, consistent with working as the left and right with high reliability
Property numerical value equal to or less than threshold value time, left and right concordance reliability values is calculated as " 1/ select left and right
Concordance numerical value " value.
17. equipment according to claim 14, wherein:
This parallax enhancement unit farther includes: left and right concordance value selector, for for first
Each pixel that each corresponding point of initial parallax image are corresponding, determines corresponding about first concordance numerical value
With the smaller value of corresponding about second concordance numerical value as having the left and right concordance numerical value of high reliability,
With
This left and right consistency checker, according to the difference of the absolute value of the parallax value for each corresponding point, is counted
Calculate the about first concordance numerical value and the about second concordance numerical value.
18. equipment according to claim 13, wherein:
This parallax enhancement unit calculates the parallax value of each neighbor as the first initial parallax figure
The weighted value of each neighbor in Xiang, or use each adjacent picture in the first initial parallax image
The parallax value of element and first strengthens the parallax value of each corresponding adjacent pixel in anaglyph, calculates every
The weighted value of one neighbor.
19. equipment according to claim 13, wherein:
This parallax enhancement unit uses between each neighbor and pixel to be strengthened and center pixel
Space length and aberration calculate the weighted value of each neighbor, or use for sectional perspective coupling
The cost sum function used is to calculate the weighted value of each neighbor.
20. equipment according to claim 13, wherein:
This parallax value determiner is for each picture corresponding with each corresponding point of the second initial parallax image
Element, use among corresponding about first concordance numerical value and corresponding about second concordance numerical value, have
The left and right concordance numerical value of higher reliability and the weighted value for each corresponding adjacent pixel, come really
Surely the parallax value of each pixel of the second initial parallax image it is used for.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020150073252A KR102324605B1 (en) | 2015-05-26 | 2015-05-26 | Method and apparatus for generating disparity image |
| KR10-2015-0073252 | 2015-05-26 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106210691A true CN106210691A (en) | 2016-12-07 |
| CN106210691B CN106210691B (en) | 2018-05-08 |
Family
ID=57453249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610357234.0A Expired - Fee Related CN106210691B (en) | 2015-05-26 | 2016-05-26 | Method and apparatus for generating anaglyph |
Country Status (2)
| Country | Link |
|---|---|
| KR (1) | KR102324605B1 (en) |
| CN (1) | CN106210691B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110463194A (en) * | 2017-03-27 | 2019-11-15 | 索尼公司 | Image processing apparatus and image processing method and image capture apparatus |
| CN112005273A (en) * | 2018-03-14 | 2020-11-27 | 谷歌有限责任公司 | Hierarchical disparity hypothesis generation with tilt-supported windows |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116630654B (en) * | 2023-06-06 | 2024-09-27 | 绿萌科技股份有限公司 | Fruit and vegetable image color detection method based on color value grade proximity weighting condition |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102074014A (en) * | 2011-02-23 | 2011-05-25 | 山东大学 | Stereo matching method by utilizing graph theory-based image segmentation algorithm |
| CN102164293A (en) * | 2010-02-16 | 2011-08-24 | 索尼公司 | Image processing device, image processing method, image processing program, and imaging device |
| CN103366354A (en) * | 2012-03-27 | 2013-10-23 | 富士通株式会社 | Method and system for stereo matching |
| US20140219549A1 (en) * | 2013-02-01 | 2014-08-07 | Electronics And Telecommunications Research Institute | Method and apparatus for active stereo matching |
| CN104079912A (en) * | 2013-03-29 | 2014-10-01 | 索尼公司 | Image processing apparatus and image processing method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101929042B1 (en) * | 2012-10-30 | 2018-12-14 | 엘지디스플레이 주식회사 | Disparity calculation unit and stereoscopic image display device including the same and disparity calculation method |
| KR101953315B1 (en) * | 2012-12-07 | 2019-02-28 | 엘지디스플레이 주식회사 | Disparity calculation method and stereoscopic image display device |
-
2015
- 2015-05-26 KR KR1020150073252A patent/KR102324605B1/en active IP Right Grant
-
2016
- 2016-05-26 CN CN201610357234.0A patent/CN106210691B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102164293A (en) * | 2010-02-16 | 2011-08-24 | 索尼公司 | Image processing device, image processing method, image processing program, and imaging device |
| CN102074014A (en) * | 2011-02-23 | 2011-05-25 | 山东大学 | Stereo matching method by utilizing graph theory-based image segmentation algorithm |
| CN103366354A (en) * | 2012-03-27 | 2013-10-23 | 富士通株式会社 | Method and system for stereo matching |
| US20140219549A1 (en) * | 2013-02-01 | 2014-08-07 | Electronics And Telecommunications Research Institute | Method and apparatus for active stereo matching |
| CN104079912A (en) * | 2013-03-29 | 2014-10-01 | 索尼公司 | Image processing apparatus and image processing method |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110463194A (en) * | 2017-03-27 | 2019-11-15 | 索尼公司 | Image processing apparatus and image processing method and image capture apparatus |
| CN112005273A (en) * | 2018-03-14 | 2020-11-27 | 谷歌有限责任公司 | Hierarchical disparity hypothesis generation with tilt-supported windows |
Also Published As
| Publication number | Publication date |
|---|---|
| KR102324605B1 (en) | 2021-11-10 |
| KR20160138847A (en) | 2016-12-06 |
| CN106210691B (en) | 2018-05-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103106651B (en) | Method for obtaining parallax error plane based on three-dimensional hough | |
| CN104820991B (en) | A kind of multiple soft-constraint solid matching method based on cost matrix | |
| CN106815594A (en) | Solid matching method and device | |
| EP1865729A2 (en) | Method and device for generating a disparity map from stereo images and stereo matching method and device therefor | |
| CN111105452B (en) | Binocular vision-based high-low resolution fusion stereo matching method | |
| CN106530333A (en) | Hierarchy optimization stereo matching method based on binding constraint | |
| Buades et al. | Reliable multiscale and multiwindow stereo matching | |
| CN106530336A (en) | Stereo matching algorithm based on color information and graph-cut theory | |
| CN106210691A (en) | For the method and apparatus generating anaglyph | |
| CN102447917A (en) | Three-dimensional image matching method and equipment thereof | |
| CN104915927A (en) | Parallax image optimization method and apparatus | |
| CN104331890B (en) | A kind of global disparity method of estimation and system | |
| CN113989758A (en) | Anchor guide 3D target detection method and device for automatic driving | |
| CN105335952A (en) | Matching cost calculation method and apparatus, and parallax value calculation method and equipment | |
| CN114445473B (en) | Stereo matching method and system based on deep learning operator | |
| CN109410235A (en) | The method for tracking target of combination of edge feature | |
| CN104408694A (en) | Denoising method for nonlocal average image based on soft threshold | |
| CN103337064A (en) | Method for removing mismatching point in image stereo matching | |
| CN109087344A (en) | Image-selecting method and device in three-dimensional reconstruction | |
| CN114241444A (en) | Lane line recognition method and apparatus, storage medium, and electronic apparatus | |
| Zhang et al. | Efficient disparity calculation based on stereo vision with ground obstacle assumption | |
| CN107122782A (en) | A kind of half intensive solid matching method in a balanced way | |
| EP3327481B1 (en) | Method for modelling an imaging device, corresponding computer program product and computer-readable carrier medium | |
| CN113344989B (en) | NCC and Census minimum spanning tree aerial image binocular stereo matching method | |
| Farid et al. | Edges shape enforcement for visual enhancement of depth image based rendering |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180508 |