CN103533375B - Video compression method and video compression device - Google Patents
Video compression method and video compression device Download PDFInfo
- Publication number
- CN103533375B CN103533375B CN201210276252.8A CN201210276252A CN103533375B CN 103533375 B CN103533375 B CN 103533375B CN 201210276252 A CN201210276252 A CN 201210276252A CN 103533375 B CN103533375 B CN 103533375B
- Authority
- CN
- China
- Prior art keywords
- vector
- motion
- block
- pixels
- current macro
- 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
- 238000000034 method Methods 0.000 title claims abstract description 19
- 230000006835 compression Effects 0.000 title claims abstract description 14
- 238000007906 compression Methods 0.000 title claims abstract description 14
- 239000013598 vector Substances 0.000 claims abstract description 166
- 238000001514 detection method Methods 0.000 claims description 16
- 239000000284 extract Substances 0.000 claims description 6
- 241000208340 Araliaceae Species 0.000 claims 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 2
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 2
- 235000008434 ginseng Nutrition 0.000 claims 2
- 230000006399 behavior Effects 0.000 claims 1
- 238000011156 evaluation Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 239000000872 buffer Substances 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 229910003460 diamond Inorganic materials 0.000 description 2
- 239000010432 diamond Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A video compression method and apparatus. The video compression method comprises the following steps: extracting a plurality of reference motion vectors of a current macro block, wherein the reference motion vectors are provided by pixel blocks with at least two sizes in a reference macro block of the current macro block; identifying a median motion vector from the reference motion vectors, and setting a search starting point of a similar macro block according to the median motion vector; searching a most similar macro block of the current macro block at the beginning of a reference frame according to the similar macro block searching starting point; and obtaining a motion vector for video compression according to the searched most similar macro block and the current macro block.
Description
Technical field
Disclosed herein a kind of video-frequency compression method and video compress device.
Background technology
Video compress includes an inter prediction (inter prediction) technology.Inter prediction includes: seek
Go out the change between a present frame (current frame) and a reference frame (reference frame) (such as,
Motion-vector/motion vector), estimate a prediction frame further according to motion-vector from this reference frame.
Prediction frame can make difference (residues) the data volume pole between this present frame and this prediction frame accurately
Little-video compress of being discussed is i.e. to compress this difference.
As shown in the above, the effect of video compress depends on the accuracy predicting frame, and predicts frame
Accuracy affected by motion-vector estimating techniques again.Motion-vector about the pattern to be encoded is estimated
Calculate, generally need first in reference frame, to search similar pattern.In reference frame, the search about similar pattern rises
The setting of point may appreciable impact video compress effect.
Summary of the invention
A kind of video-frequency compression method and device.
Include according to the video-frequency compression method that a kind of embodiment is realized: extract a current macro many
Individual with reference to motion-vector, these reference motion-vectors are by least two in the reference macroblock of this current macro
The block of pixels of size is provided;A median motion-vector is picked out with reference to motion-vector from these,
And set a similar macro blocks search starting point according to this median motion-vector;Search according to this similar macro blocks
Starting point is in a most like macro block of reference frame this current macro of initial search;And according to searching
This most like macro block and this current macro, try to achieve the motion-vector of this current macro for video compress.
Include according to the video compress device that a kind of embodiment realizes: a memory element and is moved
Dynamic detecting module.It is many that this memory element stores a current macro in a reference frame and a present frame
Individual with reference to motion-vector.These reference motion-vectors are by least two in the reference macroblock of this current macro
The block of pixels of size is provided.This mobile detection module extracts this current macro from this memory element
These are with reference to motion-vector, and pick out a median motion-vector from these with reference to motion-vector, with
Set a similar macro blocks according to this median motion-vector and search starting point.Rise according to this similar macro blocks search
Point, this this current macro of reference frame initial search that this mobile detection module is stored in this memory element
A most like macro block, and this most like macro block and current macro according to searching try to achieve motion-vector
For video compress.
For making the above-mentioned purpose of the present invention, feature and advantage to become apparent, special embodiment below,
And coordinate appended accompanying drawing, describe in detail as follows.
Accompanying drawing explanation
Fig. 1 is with a kind of embodiment of block diagram illustration video compress device;
Fig. 2 illustrates similar macro blocks and searches the definition of starting point (Initial Search Point, ISP);
Fig. 3 illustrates current macro CurMB mobile detection setting similar macro blocks and searches starting point ISP time institute
The information needed;
Fig. 4 illustrates the limit division of a present frame;
With flow chart, Fig. 5 illustrates in disclosed image compressing method that the movement about a current macro is detectd
Survey;
Fig. 6 A, Fig. 6 B, Fig. 6 C illustrate " three step searches (three step search) ", " Pedicellus et Pericarpium Trapae respectively
Shape searches (diamond search) " and " hexagonal search (hexagonal search) " technology;
Fig. 7 illustrates a median vector estimation design Median-of-4;
Fig. 8 illustrates a median vector estimation design Median-of-6;And
Fig. 9 illustrates a kind of method for searching, is used in the local search region 206 of Fig. 2 according to path
Most like macro block BestMatchedMB is sought out in instruction 902.
[main element symbol description]
102~mobile detection module;104~memory element;
106~motion compensation module;108~prediction frame;
110~two differences of interframe;
202~present frame;204~reference frame;
206~local search region;
402~first, present frame upper left corner macro block;
404~the upper edge area of present frame;
406~the left edge area of present frame;
408~the right edge area of present frame;
410~the non-edge area of present frame;
702,704~numerical value A ... two median person in D;
802,804,806,808~numerical value A ... four median person in F;
810,812~numerical value 802,804,806,808 in two median persons;
902~path instruction;
A ... F~numerical value;
Cmp1, cmp2~comparator;
Cycle1 ... Cycle5~first ... the 5th takes turns operation;
ComMB~the macro block along path instruction 902 sampling;
CurMB~current macro;
ISP~similar macro blocks search starting point;
Reference macroblock on the left of Left~current macro CurMB;
BestMatchedMB~most like macro block;
The median vector estimation design of Median-of-4~four input;
The median vector estimation design of Median-of-6~six input;
MV_L16×16、MV_L_R8×16、MV_L_D16×8、MV_L_U16×8、
MV_U16 × 16, MV_U_D16 × 8, MV_UR16 × 16, MV_UR_D16 × 8~reference are mobile
Vector;
Reference macroblock above Upper~current macro CurMB;
UpperRight~the top-right reference macroblock of current macro CurMB.
Detailed description of the invention
Fig. 1, with a kind of embodiment of block diagram illustration video compress device, wherein applies inter prediction
(inter-prediction) and two kinds of technology of infra-frame prediction (intra-prediction), the most especially
Inter-frame prediction techniques is described.
As it can be seen, a mobile detection (motion estimation) module 102 is from a memory element
104 extract about a present frame and the information of a reference frame, to make motion-vector (motion vector)
Estimation.The output of mobile detection module 102 can be supplied to a motion compensation (motion compensation)
Module 106, is allowed to the information of this reference frame is made motion compensation, it is achieved inter prediction.This moves benefit
Repay a prediction frame 108 produced by module 106 to make comparisons with present frame, difference between the two
(residues) 110 converted, quantify after, can the most scanned sequence (reorder) and entropy compile
Code (entropy encoder) is sent to network abstraction layer (NAL).This is moved by disclosed technology detects
Survey module 102 make particular design, wherein to similar macro blocks search starting point (Initial Search Point,
Being set with ISP) is specifically designed.
Content shown in Fig. 1 can realize in system single chip (Soc) mode or by computer system/micro-
Controller performs program and realizes, or, it is also possible to hardware circuit mode realizes being made as chip.
Fig. 2 illustrates and defines similar macro blocks search starting point (ISP).In present frame 202, currently it is intended to
The macro block (referred to as current macro) of coding is denoted as CurMB.So-called mobile detection is from a reference frame
204 seek out the most like macro block BestMatchedMB of most like with current macro CurMB.?
Regional area is searched in the embodiment of (partial search), need to set a similar macro blocks search and rise
Point ISP, from this similar macro blocks search that starting point ISP initiates in a local search region 206 similar grand
Block search operates.If similar macro blocks is searched starting point ISP and is set mistake, corresponding local search district
Territory will not include this most like macro block BestMatchedMB, and the effect of mobile detection can be greatly affected.
According to disclosed technology, it will be more accurate that similar macro blocks searches the setting of starting point ISP.
About current macro CurMB, Fig. 3 illustrates the similar macro blocks of its mobile detection and searches starting point
Information required for ISP setting.Figure defines multiple reference macroblocks for current macro CurMB, including:
It is positioned at the reference macroblock Left on the left of current macro CurMB;It is positioned at above current macro CurMB
Reference macroblock Upper;And it is positioned at current macro CurMB top-right reference macroblock UpperRight.
Reference macroblock Left can have multiple block of pixels dividing mode, its motion-vector (motion
Vectors) the reference motion-vector of this current macro CurMB it is used as, including:
● reference motion-vector MV_L16 × 16 that the block of pixels of 16 × 16 sizes is provided;
● the reference motion-vector that the block of pixels of right-hand part 8 × 16 size is provided
MV_L_R8×16;
● the reference motion-vector that the block of pixels of lower half 16 × 8 size is provided
MV_L_D16×8;And
● the reference motion-vector that the block of pixels of the first half 16 × 8 size is provided
MV_L_U16×8。
Reference macroblock Upper can have multiple block of pixels dividing mode, and its motion-vector is used as this
The reference motion-vector of current macro CurMB, including:
● reference motion-vector MV_U16 × 16 that the block of pixels of 16 × 16 sizes is provided;With
And
● the reference motion-vector that the block of pixels of lower half 16 × 8 size is provided
MV_U_D16×8。
● reference macroblock UpperRight can have multiple block of pixels dividing mode, its motion-vector
It is used as the reference motion-vector of this current macro CurMB, including:
● reference motion-vector MV_UR16 × 16 that the block of pixels of 16 × 16 sizes is provided;With
And
● the reference motion-vector that the block of pixels of lower half 16 × 8 size is provided
MV_UR_D16×8。
Also can be temporarily stored in above with reference to motion-vector in Fig. 1 memory element 104 for mobile detection module
102 extract use.
Different in the position of a present frame according to current macro CurMB, its mobile detection can be joined
According to reference motion-vector quantity there may be variation.Illustrate referring to Fig. 4.
Situation one, current macro CurMB is first, present frame upper left corner macro block 402.Now, when
The motion-vector (motion vector) of front macro block CurMB may be set to 0 value.
Situation two, current macro CurMB is positioned at the upper edge area 404 of present frame.Now, currently
Macro block CurMB can left side macro block (macro block that Fig. 3 label Left is illustrated) make reference macroblock.
A kind of embodiment is to make the mobile detection of this current macro CurMB can refer to defined in Fig. 3 four
Individual with reference to motion-vector: MV_L16 × 16, MV_L_R8 × 16, MV_L_D16 × 8 and
MV_L_U16×8。
Situation three, current macro CurMB is positioned at the left edge area 406 of present frame.Now, currently
Macro block CurMB can with above it with top-right macro block (Fig. 3 label Upper and UpperRight
The macro block illustrated) make reference macroblock.A kind of embodiment is the shifting making this current macro CurMB
Dynamic detecting can refer to defined in Fig. 3 four with reference to motion-vector: MV_U16 × 16,
MV_U_D16 × 8, MV_UR16 × 16 and MV_UR_D16 × 8.
Situation four, current macro CurMB is positioned at the right edge area 408 of present frame.Now, currently
Macro block CurMB can (Fig. 3 label Left Yu Upper be illustrated with macro block with top on the left of it
Macro block) make reference macroblock.A kind of embodiment is to make the mobile detection of this current macro CurMB can
With reference to defined in Fig. 3 four with reference to motion-vector: MV_L16 × 16, MV_L_R8 × 16,
MV_U16 × 16 and MV_U_D16 × 8.
Situation five, current macro CurMB is positioned at the non-edge area 410 of present frame.Now, currently
Macro block CurMB can with on the left of it, top and top-right macro block (Fig. 3 label Left, Upper
The macro block illustrated with UpperRight) make reference macroblock.A kind of embodiment is to make this current macro
The mobile detection of CurMB can refer to defined in Fig. 3 six with reference to motion-vector: MV_L16 × 16,
MV_L_R8 × 16, MV_L_D16 × 8, MV_U16 × 16, MV_U_D16 × 8 and
MV_UR16×16。
The setting that reference motion-vector design described above will make similar macro blocks search starting point ISP
To be more accurate.Special instruction, the reference motion-vector number being not intended to limit each situation described above
Measure or use situation in detail.Every pixel of at least two size in the reference macroblock of this current macro
The technology that block obtains with reference to motion-vector is the most related to the present invention.
Fig. 5 with flow chart arrange disclosed by image compressing method in about a current macro (such as Fig. 2
Label CurMB) mobile detection.Step S502, extracts the multiple with reference to mobile of this current macro
Vector;These reference motion-vectors are by the pixel of at least two size in the reference macroblock of this current macro
Block is provided-be can refer to aforementioned multiple situation to have multiple source.Step S504, from these with reference to moving
Moving vector picks out a median motion-vector, and it is similar grand to set one according to this median motion-vector
Block searches starting point (such as Fig. 2 label ISP).Step S506, searches starting point (ISP) according to this similar macro blocks
In a most like macro block of reference frame this current macro of initial search (such as Fig. 2 label
BestMatchedMB).Step S508, according to this most like macro block (BestMatchedMB) searched
Motion-vector (motion vector) with this current macro tries to achieve this current macro, makes for video compress
With.
Special declaration, above similar macro blocks search starting point (ISP) does not limit to be applied shown in Fig. 2
Regional area search technique.Three step searches (three step search) shown in Fig. 6 A, Fig. 6 B
Hexagonal search (hexagonal search) shown in shown rhombus search (diamond search), Fig. 6 C
May be used without disclosed technology and set its most like macro block search starting point ISP.
About embodiment cited by above situation two~situation four, the reference of a current macro reference is mobile
Vector quantity is 4.Fig. 7 illustrates a median vector estimation design Median-of-4, for situation two
~embodiment cited by situation four realizes Fig. 5 step S504-in four reference motion-vectors pick out
Place value motion-vector.
Median vector estimation design Median-of-4 includes two comparator cmp1 and cmp2,
Wherein recycle this two comparator cmp1 and cmp2, to complete the size of multiple motion-vector
Sequence.Each comparator (cmp1 or cmp2) is responsible for comparing the size of two inputs, and with one first
Outfan exports the greater and with one second outfan output smaller.As it is shown in fig. 7, first
Wheel operation Cycle1 in, comparator cmp1 pick out the greater Max in two input A Yu B (A,
B) with smaller Min (A, B), and comparator cmp2 picks out in two input C with D bigger
Person Max (C, D) and smaller Min (C, D).Second takes turns operation Cycle2 can reuse
Comparator cmp1 and cmp2: numerical value Max (A, B) and numerical value Max (C, D) make comparisons again,
Try to achieve maximum person Max (A, B, C, D) and one first median person 702 in numerical value A~D;
Further, numerical value Min (A, B) and numerical value Min (C, D) makes comparisons again, tries to achieve numerical value A~D
Middle minima person Min (A, B, C, D) and one second median person 704.This first and should
Corresponding to second median 702 and 704 two averagely can be considered institute with reference to the vector of motion-vector
Median motion-vector-reference step S504 of exposure, is used for setting above-mentioned similar macro blocks and searches starting point
(ISP)。
About embodiment cited by above situation five, the reference motion-vector quantity of current macro reference is
6.Fig. 8 illustrates a median vector estimation design Median-of-6, for for cited by situation five
Embodiment realize Fig. 5 step S504-from six with reference to motion-vectors pick out median move to
Amount.
Median vector estimation design Median-of-6 is to recycle two comparator cmp1 equally
And cmp2 completes the size sequence of multiple motion-vector.Operation is taken turns in the first round and second
In Cycle1 Yu Cycle2, comparator cmp1 Yu cmp2 adopts the median vector estimation design of Fig. 7
Median-of-4 operates, and picks out maximum person Max (A:D) in numerical value A~D, minima
Person Min (A:D) and two median persons 802,804.Third round operation Cycle3 reuses ratio
Relatively device cmp1 with cmp2: numerical value Max (A:D) compares with numerical value E, draws the greater Max
(A:D, E) and smaller 806;Further, numerical value Min (A:D) compares with numerical value F,
Go out smaller Min (A:D, F) and the greater 808.Then, fourth round takes turns operation Cycle4 with the 5th
With Cycle5 reuse median vector estimation design Median-of-4, pick out input 802,
804, in the middle of between two medians 810 and 812-namely six numerical value A~F in 806 and 808
Two numerical value of size.The vector of the two reference motion-vectors that median 810 is corresponding with 812 averagely may be used
It is considered disclosed median motion-vector.With reference to step S504, it is used for setting above-mentioned similar macro blocks
Search starting point (ISP).
Fig. 7 Yu Fig. 8 is not intended to limit median vector and estimates in the design shown in figure.Any
All can be applied mechanically with the design of identification median vector.
It addition, the quantity with reference to motion-vector does not limit the above embodiment with source.Below
As a example by quantity.About the reference motion-vector-such as 3 of negligible amounts, user can make Fig. 7's
One of the Median-of-4 input of estimation design of median vector is maximum value or minimum value, with should
Design Median-of-4 is modified to sort three inputs.Or, if with reference to motion-vector number
Be 5, then user can make the median vector estimation design one of Median-of-6 of Fig. 8 defeated
Enter for maximum value or minimum value, to be modified to this design Median-of-6 five inputs are sorted.
Fig. 9 illustrates a kind of method for searching, is used in the local search region 206 of Fig. 2 according to path
Most like macro block BestMatchedMB is sought out in instruction 902.
A kind of embodiment is definition one evaluation of estimate cost_value, wherein,
Cost_value=SAD+f (QP) MV,
SAD be pixel information absolute difference and, f (QP) is a function of quantization parameter QP,
MV is that the bit of motion-vector represents.
Various sizes along the macro block ComMB acquired by path instruction 902 divide (such as, one
16 × 16 block of pixels or 16 × 8 block of pixels of the first half or 16 × 8 block of pixels of lower half,
Or 8 × 16 block of pixels of left side or 8 × 16 block of pixels of right-hand part or the 4x4 in 16th district
Block of pixels or 8 × 8 block of pixels in 4th district or the 4x8 block of pixels in 8th district or the 8 × 4 of 8th district
Block of pixels) all need to carry out upper evaluation values cost_value estimation.A kind of embodiment is that pin is different
Size divides provides exclusive comparator, makes macro block belonging to each size minimum evaluation of estimate cost_value person
Motion-vector is noted down in both the buffers.Minimum evaluation of estimate cost_value of each size each other can be again
Make comparisons;Based on the motion-vector of storage in the buffer corresponding to minimum person, phase can be sought obtaining
Like macro block BestMatchedMB.
Although the present invention is disclosed above with multiple embodiments, so it is not intended to limit the present invention, appoints
What those skilled in the art, without departing from the spirit and scope of the present invention, when doing trickle change
With modification, therefore protection scope of the present invention is when being as the criterion depending on as defined in claim.
Claims (14)
1. a video-frequency compression method, including:
Extracting the multiple with reference to motion-vector of a current macro, these are the grandest by this with reference to motion-vector
In the reference macroblock of block, the block of pixels of at least two size is provided, the described reference of this current macro
Macro block is positioned on a present frame at this current macro place;
Pick out a median motion-vector from these with reference to motion-vector, and move according to this median
Vector sets a similar macro blocks and searches starting point;
Starting point is searched in a phase of reference frame this current macro of initial search according to this similar macro blocks
Like macro block;And
According to this most like macro block searched and this current macro, try to achieve the movement of this current macro to
Amount is for video compress.
Video-frequency compression method the most according to claim 1, wherein, when this current macro is positioned at
During the upper edge area of this present frame, the reference macroblock of this current macro be positioned at the left of this current macro,
And be with the block of pixels of wherein one 16 × 16 sizes provide one first with reference to motion-vector, one 8 × 16
The block of pixels of size provides one second with reference to motion-vector and the pixel region of two 16 × 8 sizes
Block provides one the 3rd and one the 4th to be somebody's turn to do with reference to motion-vector, the block of pixels next-door neighbour of this 8 × 16 size
Current macro, and the block of pixels of above-mentioned two 16 × 8 size forms this reference macroblock.
Video-frequency compression method the most according to claim 1, wherein, when this current macro is positioned at
During the left edge area of this present frame, the reference macroblock of this current macro include one first reference macroblock with
And one second reference macroblock, this first reference macroblock is positioned at the top of this current macro and is with wherein
The block of pixels of one 16 × 16 sizes provides one first with reference to motion-vector and one 16 × 8 sizes
Block of pixels provides one second reference motion-vector, and this second reference macroblock is positioned at the right side of this current macro
Top and be with the block of pixels of wherein one 16 × 16 sizes provide one the 3rd with reference to motion-vector, with
And one 16 × 8 the block of pixels of size provide one the 4th with reference to motion-vector, in this first reference macroblock
The block of pixels of this 16 × 8 size be close in this current macro, and this second reference macroblock should
The block of pixels of 16 × 8 sizes belongs to the lower half of this second reference macroblock.
Video-frequency compression method the most according to claim 1, wherein, when this current macro is positioned at
During the right edge area of this present frame, the reference macroblock of this current macro include one first reference macroblock with
And one second reference macroblock, this first reference macroblock is positioned at the left side of this current macro and is with wherein
The block of pixels of one 16 × 16 sizes provides one first with reference to motion-vector and one 8 × 16 sizes
Block of pixels provides one second reference motion-vector, and this second reference macroblock is positioned at the upper of this current macro
Side and be provide one the 3rd with reference to motion-vector with the block of pixels of wherein one 16 × 16 sizes and
The block of pixels of one 16 × 8 sizes provides one the 4th reference motion-vector, in this first reference macroblock
The block of pixels of this 8 × 16 size be close in this current macro, and this second reference macroblock this 16 × 8
The block of pixels of size is close to this current macro.
Video-frequency compression method the most according to claim 1, wherein, when this current macro is positioned at
During the non-edge area of this present frame, the reference macroblock of this current macro include one first reference macroblock,
One second reference macroblock and one the 3rd reference macroblock, this first reference macroblock is positioned at this current macro
Left side and be with the block of pixels of wherein one 16 × 16 sizes provide one first with reference to motion-vector, one
The block of pixels of 8 × 16 sizes provides one second with reference to motion-vector and the pixel of one 16 × 8 sizes
Block provides one the 3rd with reference to motion-vector, this second reference macroblock be positioned at this current macro top,
And be to provide one the 4th with reference to motion-vector, Yi Jiyi with the block of pixels of wherein one 16 × 16 sizes
The block of pixels of 16 × 8 sizes provides one the 5th reference motion-vector, and the 3rd reference macroblock is positioned at and deserves
The upper right side of front macro block and be to provide one the 6th with reference to moving with the block of pixels of wherein one 16 × 16 sizes
Moving vector, the block of pixels of this 8 × 16 size in this first reference macroblock is close to this current macro,
The block of pixels of this 16 × 8 size in this first reference macroblock belongs to the lower half of this first reference macroblock
The block of pixels of this 16 × 8 size in portion, and this second reference macroblock is close to this current macro.
6. a video compress device, including:
One memory element, stores multiple references of a current macro of a reference frame and a present frame
Motion-vector, these reference motion-vectors are by least two size in the reference macroblock of this current macro
Block of pixels is provided, and the described reference macroblock of this current macro is positioned at deserving of this current macro place
On front frame;And
One mobile detection module, extracts these reference motion-vectors of this current macro from this memory element,
And from these with reference to motion-vectors pick out a median motion-vector with according to this median move to
Amount sets a similar macro blocks and searches starting point, and searches starting point in this memory element institute according to this similar macro blocks
One most like macro block of this reference frame this current macro of initial search of storage, and should according to search
Most like macro block and this current macro try to achieve the motion-vector of this current macro for video compress.
Video compress device the most according to claim 6, wherein, when this current macro is positioned at
During the upper edge area of this present frame, the reference macroblock of this current macro be positioned at the left of this current macro,
And be with the block of pixels of wherein one 16 × 16 sizes provide one first with reference to motion-vector, one 8 × 16
The block of pixels of size provides one second with reference to motion-vector and the pixel region of two 16 × 8 sizes
Block provides one the 3rd and one the 4th to be somebody's turn to do with reference to motion-vector, the block of pixels next-door neighbour of this 8 × 16 size
Current macro, and the block of pixels of above-mentioned two 16 × 8 size forms this reference macroblock.
Video compress device the most according to claim 7, also includes:
One first comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
One second comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
Wherein:
In one the first round operate, above-mentioned first to fourth with reference to motion-vector be supplied to two-by-two this first with
And this second comparator makes comparisons;
Take turns operation in one second, this first and this second comparator again used, therebetween, this is years old
The signal that this first outfan of one comparator was exported when this first round operates is second to compare with this
The signal that this first outfan of device was exported when this first round operates make comparisons to pick out this first
To the 4th with reference to maximum person in motion-vector and one first median person, and this first comparator should
The signal that second outfan was exported when this first round operates is that this is second defeated with this second comparator
Go out to hold the signal exported when this first round operates to make comparisons to pick out this first to the 4th ginseng
Examine minima person and one second median person in motion-vector;And
This first to the 4th with reference in motion-vector recognized for this first with this second median person
The average i.e. above-mentioned median motion-vector of vector.
Video compress device the most according to claim 6, wherein, when this current macro is positioned at
During the left edge area of this present frame, the reference macroblock of this current macro include one first reference macroblock with
And one second reference macroblock, this first reference macroblock is positioned at the top of this current macro and is with wherein
The block of pixels of one 16 × 16 sizes provides one first with reference to motion-vector and one 16 × 8 sizes
Block of pixels provides one second reference motion-vector, and this second reference macroblock is positioned at the right side of this current macro
Top and be with the block of pixels of wherein one 16 × 16 sizes provide one the 3rd with reference to motion-vector, with
And one 16 × 8 the block of pixels of size provide one the 4th with reference to motion-vector, in this first reference macroblock
The block of pixels of this 16 × 8 size be close in this current macro, and this second reference macroblock should
The block of pixels of 16 × 8 sizes belongs to the lower half of this second reference macroblock.
Video compress device the most according to claim 9, also includes:
One first comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
One second comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
Wherein:
In one the first round operate, above-mentioned first to fourth with reference to motion-vector be supplied to two-by-two this first with
And this second comparator makes comparisons;
Take turns operation in one second, this first and this second comparator again used, therebetween, this is years old
The signal that this first outfan of one comparator was exported when this first round operates and this second comparator
The signal that this first outfan was exported when this first round operates make comparisons to pick out this first to
4th with reference to maximum person in motion-vector and one first median person, and this first comparator this
The signal that two outfans were exported when this first round operates and this second outfan of this second comparator
The signal exported when this first round operates is made comparisons to pick out this first to the 4th reference and is moved
Minima person and one second median person in moving vector;And
This first to the 4th with reference in motion-vector recognized for this first with this second median person
The average i.e. above-mentioned median motion-vector of vector.
11. video compress devices according to claim 6, wherein, when this current macro is positioned at
During the right edge area of this present frame, the reference macroblock of this current macro include one first reference macroblock with
And one second reference macroblock, this first reference macroblock is positioned at the left side of this current macro and is with wherein
The block of pixels of one 16 × 16 sizes provides one first with reference to motion-vector and one 8 × 16 sizes
Block of pixels provides one second reference motion-vector, and this second reference macroblock is positioned at the upper of this current macro
Side and be provide one the 3rd with reference to motion-vector with the block of pixels of wherein one 16 × 16 sizes and
The block of pixels of one 16 × 8 sizes provides one the 4th reference motion-vector, in this first reference macroblock
The block of pixels of this 8 × 16 size be close in this current macro, and this second reference macroblock this 16 × 8
The block of pixels of size is close to this current macro.
12. video compress devices according to claim 11, also include:
One first comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
One second comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
Wherein:
In one the first round operate, above-mentioned first to fourth with reference to motion-vector be supplied to two-by-two this first with
And this second comparator makes comparisons;
Take turns operation in one second, this first and this second comparator again used, therebetween, this is years old
The signal that this first outfan of one comparator was exported when this first round operates is second to compare with this
The signal that this first outfan of device was exported when this first round operates make comparisons to pick out this first
To the 4th with reference to maximum person in motion-vector and one first median person, and this first comparator should
The signal that second outfan was exported when this first round operates is that this is second defeated with this second comparator
Go out to hold the signal exported when this first round operates to make comparisons to pick out this first to the 4th ginseng
Examine minima person and one second median person in motion-vector;And
This first to the 4th with reference in motion-vector recognized for this first with this second median person
The average i.e. above-mentioned median motion-vector of vector.
13. video compress devices according to claim 6, wherein, when this current macro is positioned at
During the non-edge area of this present frame, the reference macroblock of this current macro include one first reference macroblock,
One second reference macroblock and one the 3rd reference macroblock, this first reference macroblock is positioned at this current macro
Left side and be with the block of pixels of wherein one 16 × 16 sizes provide one first with reference to motion-vector, one
The block of pixels of 8 × 16 sizes provides one second with reference to motion-vector and the pixel of one 16 × 8 sizes
Block provides one the 3rd with reference to motion-vector, this second reference macroblock be positioned at this current macro top,
And be to provide one the 4th with reference to motion-vector, Yi Jiyi with the block of pixels of wherein one 16 × 16 sizes
The block of pixels of 16 × 8 sizes provides one the 5th reference motion-vector, and the 3rd reference macroblock is positioned at and deserves
The upper right side of front macro block and be to provide one the 6th with reference to moving with the block of pixels of wherein one 16 × 16 sizes
Moving vector, the block of pixels of this 8 × 16 size in this first reference macroblock is close to this current macro,
The block of pixels of this 16 × 8 size in this first reference macroblock belongs to the lower half of this first reference macroblock
The block of pixels of this 16 × 8 size in portion, and this second reference macroblock is close to this current macro.
14. video compress devices according to claim 13, also include:
One first comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
One second comparator, compares the size of two inputs, and with one first outfan export the greater,
And with one second outfan output smaller;
Wherein:
Operating in a first round, the most above-mentioned first to the 6th with reference in motion-vector one first, 1 the
Two, one the 3rd and one the 4th vector be supplied to two-by-two this first and this second comparator make comparisons;
Take turns operation in one second, this first and this second comparator again used, therebetween, this is years old
The signal that this first outfan of one comparator was exported when this first round operates is second to compare with this
The signal that this first outfan of device was exported when this first round operates make comparisons to pick out this first
Maximum person and one first median reference to the 4th vector, and this first comparator this is second defeated
Go out to hold the signal exported when this first round operates be with this second outfan of this second comparator in
During the operation of this first round, the signal that exported makes comparisons to pick out in this first to the 4th vector
Little value person and one second median reference;
In one third round operate, this first and this second comparator again used, therebetween, above-mentioned
First to the 6th with reference to one the 5th vector in motion-vector and maximum in this first to the 4th vector
It is one the 3rd median reference that person makes comparisons to pick out smaller, and above-mentioned first to the 6th reference moves
In moving vector, one the 6th vector makes comparisons to pick out with minima person in this first to the 4th vector
The greater is one the 4th median reference;
Taking turns operation in a fourth round and one the 5th, this is first and this second comparator twice behaviour repeatedly years old
Make, to pick out one first and one second median person in above-mentioned first to fourth median reference;
And
This first to the 4th median with reference in recognized for this first with this second median person's
The average i.e. above-mentioned median motion-vector of vector.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW101124167A TWI499283B (en) | 2012-07-05 | 2012-07-05 | Video compression method and device |
TW101124167 | 2012-07-05 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103533375A CN103533375A (en) | 2014-01-22 |
CN103533375B true CN103533375B (en) | 2016-12-21 |
Family
ID=49934985
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210276252.8A Active CN103533375B (en) | 2012-07-05 | 2012-08-06 | Video compression method and video compression device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN103533375B (en) |
TW (1) | TWI499283B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2000014682A3 (en) * | 1998-09-07 | 2002-09-26 | Thomson Multimedia Sa | Method of motion estimation for transmission cost reduction of motion vectors |
CN1791224A (en) * | 2005-12-19 | 2006-06-21 | 宁波大学 | Self-adaptive block searching range rapid motion estimating method based on H.264 |
CN1829325A (en) * | 2005-03-01 | 2006-09-06 | 凌阳科技股份有限公司 | Method and system for dynamically adjusting motion estimation |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR100446083B1 (en) * | 2002-01-02 | 2004-08-30 | 삼성전자주식회사 | Apparatus for motion estimation and mode decision and method thereof |
WO2005029833A2 (en) * | 2003-09-21 | 2005-03-31 | Servision Ltd. | Deriving motion detection information from motion-vector-search type video encoders |
KR100631777B1 (en) * | 2004-03-31 | 2006-10-12 | 삼성전자주식회사 | Method and apparatus for effectively compressing motion vectors in multi-layer |
TW200727704A (en) * | 2006-01-12 | 2007-07-16 | Advance Multimedia Internet Technology Inc | Inspection method for speeding up dynamic digital image compression by reducing number of prediction modes |
TWI324484B (en) * | 2006-05-23 | 2010-05-01 | Jiunin Guo | Method for reducing calculation complexity of video compression standard |
US8488676B2 (en) * | 2007-05-14 | 2013-07-16 | Himax Technologies Limited | Motion estimation method |
US20090245374A1 (en) * | 2008-03-26 | 2009-10-01 | Mediatek Inc. | Video encoder and motion estimation method |
BR112012000002B1 (en) * | 2009-07-03 | 2021-11-23 | France Telecom | METHOD FOR PREDICTING A MOTION VECTOR OF A CURRENT IMAGE PARTITION, METHOD FOR ENCODING AN IMAGE, METHOD FOR DECODING A REPRESENTATIVE DATA FLOW OF AN IMAGE, DEVICE FOR PREDICTING A MOTION VECTOR OF A CURRENT IMAGE PARTITION, DEVICE TO ENCODE A IMAGE AND DEVICE TO DECODE A DATA FLOW |
-
2012
- 2012-07-05 TW TW101124167A patent/TWI499283B/en active
- 2012-08-06 CN CN201210276252.8A patent/CN103533375B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2000014682A3 (en) * | 1998-09-07 | 2002-09-26 | Thomson Multimedia Sa | Method of motion estimation for transmission cost reduction of motion vectors |
CN1829325A (en) * | 2005-03-01 | 2006-09-06 | 凌阳科技股份有限公司 | Method and system for dynamically adjusting motion estimation |
CN1791224A (en) * | 2005-12-19 | 2006-06-21 | 宁波大学 | Self-adaptive block searching range rapid motion estimating method based on H.264 |
Also Published As
Publication number | Publication date |
---|---|
CN103533375A (en) | 2014-01-22 |
TW201404157A (en) | 2014-01-16 |
TWI499283B (en) | 2015-09-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5555221B2 (en) | Method and apparatus for adaptive color model parameter estimation of object of interest | |
CN100468982C (en) | Method and apparatus for performing high quality fast predictive motion search | |
JP2015515806A5 (en) | ||
JP2015536092A5 (en) | ||
KR20040028911A (en) | Method and apparatus for motion estimation between video frames | |
CN102801972A (en) | Characteristic-based motion vector estimation and transmission method | |
CN102036067A (en) | Moving image encoding apparatus and control method thereof | |
KR102261669B1 (en) | Artificial Neural Network Based Object Region Detection Method, Device and Computer Program Thereof | |
US8306123B2 (en) | Method and apparatus to improve the convergence speed of a recursive motion estimator | |
Porikli | Real-time video object segmentation for MPEG-encoded video sequences | |
EP2280550A1 (en) | Mask generation for motion compensation | |
CN106331730A (en) | Double-compression detection method by using quantification factor same as H.264 video | |
Laumer et al. | Moving object detection in the H. 264/AVC compressed domain | |
CN105915916A (en) | Video steganalysis method based on motion vector rate-distortion performance estimation | |
Vass et al. | Automatic spatio-temporal video sequence segmentation | |
CN111079567B (en) | Sampling method, model generation method, video behavior identification method and device | |
CN103533375B (en) | Video compression method and video compression device | |
Michael et al. | Time series classification using compressed recurrence plots | |
Yang et al. | Spatio-temporal LBP based moving object segmentation in compressed domain | |
Takacs et al. | Feature tracking for mobile augmented reality using video coder motion vectors | |
WO2010070128A1 (en) | Method for multi-resolution motion estimation | |
KR102345258B1 (en) | Object Region Detection Method, Device and Computer Program Thereof | |
JP3711022B2 (en) | Method and apparatus for recognizing specific object in moving image | |
Liu et al. | A novel compressed domain shot segmentation algorithm on H. 264/AVC | |
US9414067B2 (en) | Methods and systems for detection of block based video dropouts |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |