CN105915835B - One kind can hard-wired frame frequency lifting method - Google Patents
One kind can hard-wired frame frequency lifting method Download PDFInfo
- Publication number
- CN105915835B CN105915835B CN201610313425.7A CN201610313425A CN105915835B CN 105915835 B CN105915835 B CN 105915835B CN 201610313425 A CN201610313425 A CN 201610313425A CN 105915835 B CN105915835 B CN 105915835B
- Authority
- CN
- China
- Prior art keywords
- motion vector
- frame
- motion
- value
- reliability
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0127—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level by changing the field or frame frequency of the incoming video signal, e.g. frame rate converter
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Television Systems (AREA)
Abstract
The invention discloses one kind can hard-wired frame frequency lifting method.It obtains motion vector predictor first with bi-directional motion estimation;Then in conjunction with time domain and spatial correlation, carries out the motion search that small diamond search is combined with local full search and obtain estimation of motion vectors value;Secondly smooth and compensation is carried out to the estimation of motion vectors value;The motion vector after smooth and compensation is modified followed by image information, obtains terminating motion vector;Finally according to the requirement of frame frequency lifting, obtain the motion vector of interpolated frame, judged according to the reliability of the interpolated movements vector, determine whether to filter, when reliability is low, final interpolation result is obtained in the method that the motion vector of the interpolated frame and filtered interpolated frame motion vectors resultants weight, generates interpolated frame.The frame frequency lifting method of the present invention using can hard-wired system architecture, improve the efficiency of algorithm, improve the accuracy and reliability of motion vector, and obtained final interpolation result accuracy is high.
Description
Technical field
The present invention relates to a kind of fields that video screen is shown, and in particular to one kind can hard-wired frame frequency lifting side
Method.
Background technology
In recent years, consumer requires to improve year by year for the image quality of liquid crystal display LCD TV, therefore LCD television screen
Size also begin to high definition even super clear TV signal occur while constantly increase.But LCD TV was responded due to it
Slow disadvantage causes the blooming of moving image.Currently, the interpolation method based on Motion estimation and compensation is used to
The frame frequency lifting for realizing HD video, to solve the problems, such as that moving object is fuzzy.
Since the frame frequency lifting method calculation amount based on pixel is excessive, frequently with the method based on macroblock match come
Realize frame frequency lifting.Frame frequency lifting includes two parts of Motion estimation and compensation.Traditional movement based on macroblock match
Estimation includes one-way movement estimation and bi-directional motion estimation.One-way movement estimation will produce the problem of overlapping and cavity;Two-way fortune
Dynamic estimation can then be influenced by search range.Therefore it is not high to obtain motion vector reliability for general method.
For motion compensation, the motion vector that is generally obtained using estimation carries out the macroblock match of front and back frame, and
Interleave is carried out using the pixel that matching obtains.The general obtained pixel value precision of motion compensation is not high, especially to movement
Complicated object hardly results in accurate interpolation result.
Invention content
[technical problems to be solved]
Present invention aim to address above-mentioned prior art problems, and providing one kind can hard-wired frame frequency lifting method.
This method be intended to hardware it is achievable in the case of it is efficient and it is high-precision reach frame frequency lifting, to meet high score under high refresh rate
The display effect of resolution LCD television screen.
[technical solution]
In order to reach above-mentioned technique effect, the present invention takes following technical scheme:
One kind can hard-wired frame frequency lifting method, it includes the following steps:Estimation of motion vectors, motion vector are flat
Sliding, motion compensation and interpolated frame generation, and entire step realizes that its specific steps are such as by way of Block- matching
Under:
(1) macro block for dividing each frame into many non-overlapping copies carries out bidirectional-movement to each macro block of adjacent two frame and estimates
Meter, obtains the motion vector predictor of former frame and each macro block of present frame;
(2) time domain and spatial correlation are combined, according to the motion vector of former frame and present frame same coordinate position macro block
Predicted value is carried out the motion search that small diamond search is combined with local full search, is obtained with different estimation matching criterior
To estimation of motion vectors value;
(3) the smooth of motion vector and compensation are carried out to obtained estimation of motion vectors value;
(4) motion vector after smooth and compensation is further repaiied using former frame and the image information of present frame
Just, the final value motion vector obtained;
(5) according to the requirement of frame frequency lifting, the motion vector of interpolated frame is obtained, according to the reliable of the interpolation frame motion vector
Property judge, it is determined whether filtering;When reliability is high, the generation of interpolated frame is carried out with the interpolation frame motion vector;Work as reliability
When low, final interpolation knot is obtained in the method that the motion vector of the interpolated frame and filtered interpolated frame motion vectors resultants weight
Fruit generates interpolated frame;When reliability is between height, laggard row interpolation is filtered to the interpolation frame motion vector
The generation of frame.
According to the present invention with further technical solution, the different estimation matching criterior be SAD matching criterior,
VOD matching criterior, MAD matching criterior, MSE matching criterior or NCCF matching criterior.
The movement being combined with further technical solution, the small diamond search and local full search according to the present invention is searched
Rope refers to equal when former frame with the motion vector predictor of present frame same coordinate position macro block and when reliability is high, with the fortune
Small diamond search is done centered on the result of dynamic vector predicted value, obtains estimation of motion vectors value;When former frame is identical as present frame
When the motion vector predictor of coordinate position macro block is unequal or reliability is low, respectively with the motion vector predictor of former frame
As a result it is carried out centered on the result of the motion vector predictor centered on, by present frame and centered on the target macroblock of present frame
Three small diamond search;When obtained value is less than preset threshold value, using the value as estimation of motion vectors value;When what is obtained
When value is not less than preset threshold value, local full search is enabled, is estimated by putting bidirectional-movement pixel-by-pixel in given range
Meter, obtains estimation of motion vectors value.
According to the present invention with further technical solution, the judgement of motion vector and motion vector predictor reliability is
According to the corresponding affiliated range of macro block corresponding pixel points pixel value, threshold value th_low and th_high;Then corresponding picture is utilized
The difference blk_dif of element and the relationship of threshold value carry out the judgement of reliability;As blk_dif≤th_low, reliability is high;Work as blk_
Dif >=th_high, reliability are low.
According to the present invention with further technical solution, the smooth and compensation of the motion vector be with medium filtering and
Mean filter is filtered operation.
According to the present invention with further technical solution, the flat of motion vector has been carried out to obtained estimation of motion vectors value
After sliding and compensation, motion vector can also be carried out and split operation and global vector statistics.
According to the present invention with further technical solution, the utilization image information refers to utilizing corresponding macro block respective pixel
The pixel value of point judges the reliability of corresponding macro block final value motion vector;The low final value motion vector of reliability is repaiied
Just.
According to the present invention with further technical solution, the motion vector of the interpolated frame be before being calculated from interpolation phase to
Motion vector MV_P and backward motion vector MV_N;The calculation formula of the MV_P and MV_N is as follows:
MV_P=MV*phase/32
MV_N=MV* (64-phase)/32
Wherein, the motion vector is final value motion vector;Phase is interpolation phase.
It will be described in detail the present invention below.
Bi-directional motion estimation is carried out to each macro block of adjacent two frame, certain is counted centered on target macroblock to former frame
The standard deviation of macro block in range divides motor pattern according to the standard deviation acquired, is finally transported with same motor pattern macro block in window
Motion vector predictor of the mean value of dynamic vector as target macroblock;And for the target macroblock of present frame, by around it
Motion vector predictor of the intermediate value of calculated motion vector as target macroblock.
The preferred embodiments of the invention are that the prediction of motion vector is carried out using the pixel macroblock of 16*16 as basic unit.
Bi-directional motion estimation to former frame is that the standard deviation of motion vector is counted in the window of the 3*3 centered on target macroblock, is calculated
Former frame motion vector predictor.
The motion search for carrying out small diamond search and local full search is combined, be value pass through obtained former frame with
The result and reliability of current frame motion vectors predicted value carry out a series of motion vector that small diamond search obtain, further according to
The motion vectors resultants determine whether to carry out local full search, and the small diamond search and local full search are processes arranged side by side.
Such combined method not only ensure that the reliability of algorithm but also reduce the computation complexity of algorithm as far as possible so that algorithm is real
With property higher.
The preferred embodiment of the present invention is to carry out estimation with SAD matching criterior, is carried out down using obtained sad value
The calculating and judgement of one step.
The present invention has carried out a filtering fortune to carry out the smooth of motion vector and compensation to estimation of motion vectors value
It calculates.By taking horizontal direction X as an example, takes 8 blocks of current block and surrounding to form the motion vector block of a 3*3, calculate this 9 X-directions
The mean value of upper motion vector components, and the number that 9 components are zero is counted simultaneously;Then the mean difference for calculating 9 components, with flat
Inequality is that the X-component of thresholding and current block and the equal value difference of 9 components compare, if this difference is bigger than thresholding, it is judged that
The X-component of current block belongs to singular value;Similarly judge the singularity of the vertical direction Y-component of current block.Last Rule of judgment, such as
The horizontally or vertically component of fruit current block is there are one singular value is belonged to, then being just filtered to the motion vector of current block.Such as
Fruit needs to filter the block, then judges whether is horizontal or vertical direction component is zero in the 3*3 neighborhoods where current block number
More than 3, if meeting this condition there are one horizontal or vertical direction components, the output of current block motion vector after filtering
For the mean value of horizontal and vertical component;If condition is unsatisfactory for, filtering output is that 3*3 is filtered in horizontal and vertical component
Value.
In order to further increase the accuracy of motion vector, fractionation operation and the global vector of motion vector can also be carried out
Statistics.The fractionation operation method is as follows:The current motion vector acquired is the pixel macroblock for being directed to 16*16, in order to
Precision is improved, the macro block for being split as 8*8 is handled.4 blocks with identical vector are first divided into, it is then each
A sub-block carries out a field multilevel iudge filtering, and specific practice is to compare 3 most adjacent sub-block vectors of the sub-block vector sum
Along with the SAD of zero vector corresponding data block, minimum SAD corresponds to the final motion vector of the sub-block.Global statistics refer to complete
The motion vector that place has is counted, and judges whether that most motion vectors are identical, if it is it is considered that whole field all has
There are such motion state, referred to as static exercise.
The present invention tables look-up the organizational form for obtaining interpolated frame, by interpolation phase according to input video frame number and target frame number
(phase) forward motion vector MV_P and backward motion vector MV_N is calculated.Wherein each interpolation section is divided into 64 parts, often
The corresponding once phase of part, so referred to as 64 degree of phase accuracy interpolation.Therefore, when phase is 32 degree, motion vector is not necessarily to adjustment;
When phase is 0 degree, primitive frame is not necessarily to interpolation arithmetic.
If blk_dif<=th_low, then it is assumed that corresponding position is that reliability is high in match block, that is, it is high that motion vector is arranged
Low reliability weighted value alpha=0, if blk_dif>=th_high, then it is assumed that corresponding position is that reliability is low in match block,
That is setting motion vector height reliability weighted value alpha=1.When it falls between, the calculation of weighted value is schemed
Alpha=threshold values described in 3Computational methods of threshold value are at this:
In the present invention, when motion vector reliability is low, need the correcting mode to motion vector as follows:
Subregion is carried out according to the method for Fig. 4:Centrally located 8*8 blocks are divided into tetra- regions A, B, C and D.Center
8*8 blocks are current block, and N1~N8 is neighbours' block.Wherein the motion vector of the sub-block of tetra- 4*4 of A, B, C and D is that it is most adjacent
The result of four macroblock motion vector mean filters.For example, for sub-block A, most adjacent four macro blocks be N1, N2, N3 and
N4;For sub-block C, most adjacent four macro blocks are N4, N6, N7 and N8.
[advantageous effect]
Compared with prior art, the present invention having advantageous effect below:
The present invention, which compares existing frame frequency lifting algorithm, has following advantageous effect:Frame frequency lifting side provided by the invention
Method, using can hard-wired system architecture, improve the efficiency of algorithm;In estimation of motion vectors, time domain spy has been considered
Property and spatial character, improve the accuracy of motion vector.In addition, motion compensation has been done using mean filter and medium filtering,
Further improve the reliability of motion vector.In the interpolation part of algorithm, after the motion vector of interpolated frame and filtering
Interpolated frame motion vectors resultants weighting method obtain final interpolation result, finally obtained more accurate interpolation frame.
Description of the drawings
Fig. 1 be the present invention can hard-wired frame frequency lifting method flow diagram;
Fig. 2 is that the present invention can bi-directional motion estimation can hard-wired algorithm block diagram in hard-wired frame frequency lifting method
Schematic diagram;
Fig. 3 be the present invention can in hard-wired frame frequency lifting method motion vector smooth and compensation and interpolating module life
At algorithm block diagram schematic diagram;
Fig. 4 be the present invention can in hard-wired frame frequency lifting method motion compensation the signal of motion vector refinement definition graph
Figure.
Specific implementation mode
With reference to the embodiment of the present invention, the invention will be further elaborated.
As shown in FIG. 1 to 3, one kind can hard-wired frame frequency lifting method, it includes the following steps:
(1) macro block for dividing each frame into many non-overlapping copies carries out bidirectional-movement to each macro block of adjacent two frame and estimates
Meter, obtains the motion vector predictor of former frame and each macro block of present frame;To 3*3 of the former frame centered on target macroblock
The standard deviation of statistics motion vector in window, divides motor pattern, finally with same motor pattern in window according to the standard deviation acquired
Motion vector predictor of the mean value of the motion vector of macro block as target macroblock.And for the target macroblock of present frame, by it
Surrounding has calculated motion vector predictor of the intermediate value as target macroblock of motion vector.
(2) time domain and spatial correlation are combined, according to the motion vector of former frame and present frame same coordinate position macro block
Predicted value carries out the motion search that small diamond search is combined with local full search, obtains motion vector with SAD matching criterior and estimate
Evaluation;When former frame is equal with the motion vector predictor of present frame same coordinate position macro block and reliability is high, with the fortune
Small diamond search is done centered on the result of dynamic vector predicted value, obtains estimation of motion vectors value;When former frame is identical as present frame
When the motion vector predictor of coordinate position macro block is unequal or reliability is low, respectively with the motion vector predictor of former frame
As a result it is carried out centered on the result of the motion vector predictor centered on, by present frame and centered on the target macroblock of present frame
Three small diamond search;When obtained value is less than preset threshold value, using the value as estimation of motion vectors value;When what is obtained
When value is not less than preset threshold value, local full search is enabled, is estimated by putting bidirectional-movement pixel-by-pixel in given range
Meter, obtains estimation of motion vectors value.
Wherein, the judgement of motion vector and motion vector predictor reliability is according to corresponding macro block corresponding pixel points
The affiliated range of pixel value, threshold value th_low and th_high;Then the pass of the difference blk_dif and threshold value of respective pixel are utilized
System carries out the judgement of reliability;As blk_dif≤th_low, reliability is high;As blk_dif >=th_high, reliability is low.
(3) the smooth of motion vector and compensation are carried out to obtained estimation of motion vectors value;Be with medium filtering and
Value filtering is filtered operation.
(4) motion vector after smooth and compensation is further repaiied using former frame and the image information of present frame
Just, the final value motion vector obtained;
(5) according to the requirement of frame frequency lifting, the motion vector of interpolated frame is obtained, according to the reliable of the interpolation frame motion vector
Property judge, it is determined whether filtering;When reliability is high, the generation of interpolated frame is carried out with the interpolation frame motion vector;Work as reliability
When low, final interpolation knot is obtained in the method that the motion vector of the interpolated frame and filtered interpolated frame motion vectors resultants weight
Fruit generates interpolated frame;When reliability is between height, laggard row interpolation is filtered to the interpolation frame motion vector
The generation of frame.
Another embodiment of the invention has carried out the smooth of motion vector and compensation to obtained estimation of motion vectors value
Afterwards, motion vector can also be carried out and split operation and global vector statistics.The fractionation operation method is as follows:Currently acquire
Motion vector is the pixel macroblock for being directed to 16*16, and in order to improve precision, the macro block for being split as 8*8 is handled.First
4 blocks with identical vector are divided into, then each sub-block carries out a field multilevel iudge filtering, specific practice
It is the SAD for comparing 3 most adjacent sub-block vectors of the sub-block vector sum and adding zero vector corresponding data block, minimum SAD is corresponded to
The final motion vector of the sub-block.Global statistics refer to that the motion vector all to the whole audience counts, and judge whether most fortune
Dynamic vector is identical, if it is it is considered that whole field all has such motion state, referred to as static exercise.
Described using image information refers to judging that corresponding macro block final value is transported using the pixel value of corresponding macro block corresponding pixel points
The reliability of dynamic vector;The low final value motion vector needs of reliability are modified.
The motion vector of the interpolated frame is to calculate forward motion vector MV_P and backward motion vector MV_ by interpolation phase
N;The calculation formula of the MV_P and MV_N is as follows:
MV_P=MV*phase/32
MV_N=MV* (64-phase)/32
Wherein, the motion vector is final value motion vector;Phase is interpolation phase.
The present invention tables look-up the organizational form for obtaining interpolated frame, by interpolation phase according to input video frame number and target frame number
(phase) forward motion vector MV_P and backward motion vector MV_N is calculated.Wherein each interpolation section is divided into 64 parts, often
The corresponding once phase of part, so referred to as 64 degree of phase accuracy interpolation.Therefore, when phase is 32 degree, motion vector is not necessarily to adjustment;
When phase is 0 degree, primitive frame is not necessarily to interpolation arithmetic.
If blk_dif<=th_low, then it is assumed that corresponding position is that reliability is high in match block, that is, it is high that motion vector is arranged
Low reliability weighted value alpha=0, if blk_dif>=th_high, then it is assumed that corresponding position is that reliability is low in match block,
That is setting motion vector height reliability weighted value alpha=1.When it falls between, the calculation of weighted value is schemed
Alpha=threshold values described in 3Computational methods of threshold value are at this:
In the present invention, when motion vector reliability is low, need the correcting mode to motion vector as follows:
Subregion is carried out according to the method for Fig. 4:Centrally located 8*8 blocks are divided into tetra- regions A, B, C and D.Center
8*8 blocks are current block, and N1~N8 is neighbours' block.Wherein the motion vector of the sub-block of tetra- 4*4 of A, B, C and D is that it is most adjacent
The result of four macroblock motion vector mean filters.For example, for sub-block A, most adjacent four macro blocks be N1, N2, N3 and
N4;For sub-block C, most adjacent four macro blocks are N4, N6, N7 and N8.
Although reference be made herein to invention has been described for explanatory embodiment of the invention, and above-described embodiment is only this hair
Bright preferable embodiment, embodiment of the present invention are not limited by the above embodiments, it should be appreciated that people in the art
Member can be designed that a lot of other modification and implementations, these modifications and implementations will be fallen in principle disclosed in the present application
Within scope and spirit.
Claims (7)
1. one kind can hard-wired frame frequency lifting method, it is characterised in that it includes the following steps:Estimation of motion vectors, movement
Vector is smooth, motion compensation and interpolated frame generation, and entire step is realized by way of Block- matching, it specific
Steps are as follows:
(1) macro block for dividing each frame into many non-overlapping copies carries out bi-directional motion estimation to each macro block of adjacent two frame, obtains
To the motion vector predictor of former frame and each macro block of present frame;
(2) time domain and spatial correlation are combined, according to the motion-vector prediction of former frame and present frame same coordinate position macro block
Value is carried out the motion search that small diamond search is combined with local full search, is transported with different estimation matching criterior
Dynamic vector estimated value;
(3) the smooth of motion vector and compensation are carried out to obtained estimation of motion vectors value;
(4) it is corrected, is obtained using former frame and the image information of present frame are further to the motion vector after smooth and compensation
To final value motion vector;
(5) according to the requirement of frame frequency lifting, the motion vector of interpolated frame is obtained, is sentenced according to the reliability of the interpolation frame motion vector
It is disconnected, it is determined whether filtering;When reliability is high, the generation of interpolated frame is carried out with the interpolation frame motion vector;When reliability is low,
Final interpolation result is obtained in the method that the motion vector of the interpolated frame and filtered interpolated frame motion vectors resultants weight, it is raw
At interpolated frame;When reliability is between height, laggard row interpolation frame is filtered to the interpolation frame motion vector
It generates;
The motion vector of the interpolated frame is to calculate forward motion vector MV_P and backward motion vector MV_N by interpolation phase;Institute
The calculation formula for stating MV_P and MV_N is as follows:
MV_P=MV*phase/32
MV_N=MV* (64-phase)/32
Wherein, the MV is final value motion vector;Phase is interpolation phase.
2. according to claim 1 can hard-wired frame frequency lifting method, it is characterised in that the different movement is estimated
Meter matching criterior is SAD matching criterior, VOD matching criterior, MAD matching criterior, MSE matching criterior or NCCF matching criterior.
3. according to claim 1 can hard-wired frame frequency lifting method, it is characterised in that the small diamond search and
The motion search that local full search is combined refers to the motion-vector prediction when former frame and present frame same coordinate position macro block
Be worth equal and reliability it is high when, with small diamond search is done centered on the result of the motion vector predictor, obtain motion vector and estimate
Evaluation;When former frame with the motion vector predictor of present frame same coordinate position macro block is unequal or reliability is low when, respectively
Centered on the result of motion vector predictor centered on the result of the motion vector predictor of former frame, by present frame and with
Three small diamond search are carried out centered on the target macroblock of present frame;When obtained value is less than preset threshold value, with this
Value is estimation of motion vectors value;When obtained value is not less than preset threshold value, local full search is enabled, in given model
By putting bi-directional motion estimation pixel-by-pixel in enclosing, estimation of motion vectors value is obtained.
4. according to claim 3 can hard-wired frame frequency lifting method, it is characterised in that motion vector and movement are sweared
The judgement for measuring predicted value reliability is according to the corresponding affiliated range of macro block corresponding pixel points pixel value, threshold value th_low
And th_high;Then the judgement of reliability is carried out using the relationship of the difference blk_dif of respective pixel and threshold value;When blk_dif≤
Th_low, reliability are high;As blk_dif >=th_high, reliability is low.
5. according to claim 1 can hard-wired frame frequency lifting method, it is characterised in that the motion vector it is flat
Sliding and compensation is to be filtered operation with medium filtering and mean filter.
6. according to claim 5 can hard-wired frame frequency lifting method, it is characterised in that obtained motion vector
After estimated value has carried out the smooth of motion vector and compensation, motion vector can also be carried out and split operation and global vector statistics.
7. according to claim 4 can hard-wired frame frequency lifting method, it is characterised in that the final value motion vector
The pixel value of corresponding macro block corresponding pixel points is utilized to judge the reliability of corresponding macro block final value motion vector;Reliability low end
Value motion vector needs are modified.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2016100845469 | 2016-02-05 | ||
CN201610084546 | 2016-02-05 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105915835A CN105915835A (en) | 2016-08-31 |
CN105915835B true CN105915835B (en) | 2018-10-19 |
Family
ID=56748135
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610313425.7A Active CN105915835B (en) | 2016-02-05 | 2016-05-12 | One kind can hard-wired frame frequency lifting method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105915835B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106791279B (en) * | 2016-12-30 | 2020-01-03 | 中国科学院自动化研究所 | Motion compensation method and system based on occlusion detection |
CN108492320B (en) * | 2018-03-14 | 2022-04-12 | 四川长九光电科技有限责任公司 | Infrared dim target detection method based on parallel processing |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1416642A (en) * | 2000-12-11 | 2003-05-07 | 皇家菲利浦电子有限公司 | Motion compensated de-interlacing in video signal processing |
CN1961338A (en) * | 2004-04-09 | 2007-05-09 | 索尼株式会社 | Image processing apparatus and method, and recording medium and program |
CN1972421A (en) * | 2005-11-25 | 2007-05-30 | 三星电子株式会社 | Frame interpolator, frame interpolation method and motion credibility evaluator |
CN101543068A (en) * | 2007-03-23 | 2009-09-23 | 索尼株式会社 | Image processing device, image processing method, program, and image display device |
JP2009290828A (en) * | 2008-06-02 | 2009-12-10 | Canon Inc | Image processor, and image processing method |
CN105100807A (en) * | 2015-08-28 | 2015-11-25 | 山东大学 | Motion vector post-processing based frame rate up-conversion method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2907301A1 (en) * | 2006-10-12 | 2008-04-18 | Thomson Licensing Sas | METHOD OF INTERPOLATING A COMPENSATED IMAGE IN MOTION AND DEVICE FOR CARRYING OUT SAID METHOD |
-
2016
- 2016-05-12 CN CN201610313425.7A patent/CN105915835B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1416642A (en) * | 2000-12-11 | 2003-05-07 | 皇家菲利浦电子有限公司 | Motion compensated de-interlacing in video signal processing |
CN1961338A (en) * | 2004-04-09 | 2007-05-09 | 索尼株式会社 | Image processing apparatus and method, and recording medium and program |
CN1972421A (en) * | 2005-11-25 | 2007-05-30 | 三星电子株式会社 | Frame interpolator, frame interpolation method and motion credibility evaluator |
CN101543068A (en) * | 2007-03-23 | 2009-09-23 | 索尼株式会社 | Image processing device, image processing method, program, and image display device |
JP2009290828A (en) * | 2008-06-02 | 2009-12-10 | Canon Inc | Image processor, and image processing method |
CN105100807A (en) * | 2015-08-28 | 2015-11-25 | 山东大学 | Motion vector post-processing based frame rate up-conversion method |
Also Published As
Publication number | Publication date |
---|---|
CN105915835A (en) | 2016-08-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ha et al. | Motion compensated frame interpolation by new block-based motion estimation algorithm | |
CN1992789B (en) | Motion estimator and motion method | |
JP5645636B2 (en) | Frame interpolation apparatus and method | |
CN104219533B (en) | A kind of bi-directional motion estimation method and up-conversion method of video frame rate and system | |
US8817878B2 (en) | Method and system for motion estimation around a fixed reference vector using a pivot-pixel approach | |
CN105282475B (en) | Crawl detection and compensation method and system | |
KR20030033851A (en) | Apparatus and method for converting frame and/or field rate using adaptive motion compensation | |
EP1987676A1 (en) | Method and apparatus for determining motion between video images | |
CN103369208B (en) | Self adaptation interlace-removing method and device | |
US20090115908A1 (en) | Method and System for Automatically Turning Off Motion Compensation When Motion Vectors are Inaccurate | |
Tai et al. | A multi-pass true motion estimation scheme with motion vector propagation for frame rate up-conversion applications | |
KR20100118978A (en) | Sparse geometry for super resolution video processing | |
Heinrich et al. | Optimization of hierarchical 3DRS motion estimators for picture rate conversion | |
US20120008689A1 (en) | Frame interpolation device and method | |
CN105915835B (en) | One kind can hard-wired frame frequency lifting method | |
CN102447870A (en) | Detection method for static objects and motion compensation device | |
US8761262B2 (en) | Motion vector refining apparatus | |
CN102364933A (en) | Motion-classification-based adaptive de-interlacing method | |
CN108270945A (en) | A kind of motion compensation denoising method and device | |
CN201222771Y (en) | High speed edge self-adapting de-interlaced interpolation device | |
US20110149150A1 (en) | System, method, and apparatus for providing improved high definition video from upsampled standard definition video | |
CN105657319B (en) | The method and system of candidate vector penalty value are controlled in ME based on feature dynamic | |
US10587840B1 (en) | Image processing method capable of deinterlacing the interlacing fields | |
CN109788297B (en) | Video frame rate up-conversion method based on cellular automaton | |
US8531598B2 (en) | Image processing method and related apparatus for generating target image block |
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 |