Connect public, paid and private patent data with Google Patents Public Datasets

Derivation and use of motion vectors in a differential pulse code modulation system

Info

Publication number
CA2078950A1
CA2078950A1 CA 2078950 CA2078950A CA2078950A1 CA 2078950 A1 CA2078950 A1 CA 2078950A1 CA 2078950 CA2078950 CA 2078950 CA 2078950 A CA2078950 A CA 2078950A CA 2078950 A1 CA2078950 A1 CA 2078950A1
Authority
CA
Grant status
Application
Patent type
Prior art keywords
motion
frame
block
frames
anchor
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.)
Abandoned
Application number
CA 2078950
Other languages
French (fr)
Inventor
William E. Lynch
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Panasonic Corp
Original Assignee
William E. Lynch
Matsushita Electric Corporation Of America
Matsushita Electric Industrial Co., Ltd.
Priority date (The priority date 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 date listed.)
Filing date
Publication date

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51Motion estimation or motion compensation
    • H04N19/577Motion compensation with bidirectional frame interpolation, i.e. using B-pictures
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51Motion estimation or motion compensation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/61Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding

Abstract

ABSTRACT

A differential pulse code modulation system having bidirectional motion estimation for blocks in B
frames that is derived from the motion vector of a block in an anchor frame whose projection along its motion vector has the most overlap with the block in the B frame for which a motion estimation is sought.

Description

2~7~950 M~T~030 DE~IV~TION ~ND USE OF MOTION VECTORS IN ~ DIFFERENTIA~
PUL,SE CODE MODUL~TION SYSTEM

BACKGROUND OF THE INVENTION

In a Differential Pulse Code Modulation (DPCM) system referred to as Independent Search, IS, that has . been suggested by the Moving Picture Experts Group and-described in articles 1, 2 and 3, identified at the end of the specification that are hereby incorporated by reference, three types of frames are used as illustrated in Fig. 1. The letters indicate the type o~ frame, and the numerical subscripts indicate the order of occurrence of the frames at the input to the system. I0 and I9 are called intraframes. Each of these intraframes is encoded ;
, dlm-015 ~ mata-030~patO01 .. ...

. , ~ ' , , ' .
. . .
~, .

~i~ ' 2~78~0 M~ 030 - 2 -' , by using only pixels within the rame. The encoded frames are then transmitted.

~ predicted frame such as P3 is encoded from I~
in the following manner. P3 is partltioned into identical blocks, e.g. having eight pixels in a ~lde, that are called motion blocks. For each mo~ion block in - P3, a search is made in the I0 frame for a block of the same shape and size that matches the motion block. When the matching block is located, its po~ition relative to the motion block is indicated by a motion vector (mv) representing the number of pixels lef~ or right and the number of pixels up or down from the motion block to the matching block. These motion vectors axe transmitted.
Also, the differences between the motion blocks and corresponding matching blocks are formed. This is called ;~ residue, and it is encoded and transmitted. At the decoder, the Io frame is decoded and the predicted frame P3 is formed by displacing appropriate matching blocks in I0 using the motion vectors. The residue for the various blocks is then added. The frame P6 is encoded in like manner from P3. When a number of P frames have been encoded from preceding P frames in this manner, the accumulative error may become so great that another I
frame is sent as described, and a new series of P frames is formed from it. The I and P frames are called anchor fremes.

;' ' . . .
: . .

";

2~ 9~0 M~rA-030 - 3 -The s ~rames occur between anchor ~rames, and the motio~ vectors u~ed in encoding them are illustrated in Fig. 2. The firs~ numeral in a subscript of a motion vector mv indicates the frame in which a motion block i9 located and the second numeral indicates the frame being searched for a matching block. Thus, in encoding the frame P3 as noted above, mv30 i9 a motion vector indicating the position of a block in I0 that matches a motion bloclc in frame P3 and mv63 i9 a motion vector ~ 10 indicating the position of a block in P3 that matches a ; motion block in P6.

For each motion block in a B frame, two motion vectors are deri~ed, one referenced to a subsequent anchor frame and one anchor referenced to a previous anchor frame. Thus, mv23 indicates the position of a matching block in P3 relative to a motion block in B2 and .~
mv20 indicates the position of a matching block in I0 ~` relative to a motion bloclc in B2. What to transmit to ;~ .
the decoder can be determined by finding for each block which of the following modeg of operation produces the least Mean Square Error (MSE~ as this will reduce the bit ~ xate required. Eor each motion block, the MSE i9 ,~ calculated by adding the squa~es of the differences between the corresponding pixels of the motion block and the matching block and dividing the rasult by the number of pixels in the motion block. In what is called a backward mode, the MSE is calculated between the mo~ion ..~.

' ,: `

~' ~
, ~ .

.. ; ,, . r~ ; r ~ C.~ , S; ' "~:
r ' , ~ ~ ' .
~' ', ... .
~ .
:

.
2~7~g~0 MAT~-030 - ~ -block in s2 and the matching block in P3, and in a forward mode, the MSE calcu]ation is done with the matching bloclc being in I0. In what is called an interpolative mode, the matching blocks in P3 and I0 are averaged, and the MSE i9 calculated with this averaged matching bloclc. Whichever mode produces the least MSF is conveyed to the decoder. Thus, if the backward mode produces the least MSE, a motion vector mv23 i3 transmitted along with an indication that it is the backward vector, and if the forward mode produces the least MSE, a motion vector mv20 is tran~mitted along with an indication that it is the Eorward vector. If the interpolative mode produces the least MSE, both vectors mv23 and mv20 are transmitted along with an indication that both vectors are being transmitted. Of cour9e, for the selected mode, the re~idue obtained by subtracting the matching block from the motion block i9 also encoded and transmitted.

Fig. 1 shows the various frames in the order that they are received by the encoder and pre~ented to the viewer, and Fig. 3 illustrates the order in which the frame~ of Fig. 1 are procesged. Both P 3 and Io must be available before B-2 and B 1 can be processed as described above, and Io must be retained in memory until P3 is available before B1 and B2 can be processed. The frame order of Fig. 3 is also the order in whlch the frames are transmitted.

,, .

, 2~8~
MAT~-030 5 In sw~nary, transmission must include the Io frames, the P to I frame motion vectors such as mv30, the residue for predicted frame~ such as P3, the residue for each B frame that results from the selected mode of operation and at least one motion vector for each B frame along with signals indicating the mode of operation that has been selected. How a decoder can reconstruct the frameY from this information is thought to be obvious.

In article 4, which i~ i~corporated herein by reference, a differential pulse code modulation system i9 described that i9 like the IS system de~cribed above except for the fact that the motion vectors for the B
frames ~re derived from motion vectors such as mv30 or mv63 of Fig. 2 that indicate the relative po~ition of a block in an anchor frame that matches a block in the next anchor frame. The motion vector selected for a motlon block in a B frame is the motion vector for a block in the later anchor frame that is in the same spatial position. If linear motion i9 assumed, a motion vector mv20 (i, j) would equal 2/3 mv30 (i, j) and mv23 (i, j3 would equal -1/3 mv30 (i, j). While this method has the advantage o~ not transmitting motion vectors for the B
~rames, the motion vectors that are used re~uire the transmission of a significant amount of residue in most cases.

. ' ' . ~ ~

2~78~
MA'l`~-030 - 6 -In article 5, which is incorporated herein by reEerence, the bloclcs in khe earlier and later anchor frames that are at the ends of a vector such as mv30 or mv63 that pass through a block in a B frame are averaged to produce the data in that block. ~lthough this method doe~ not requlre the transmi~ion oE motion vectors for the B frames! it suffers from gap and overlap problems.
I
BRIEF DESCRIPTION OF THE INVENTION
In accordance with this invention, motion vectors for a block in a B frame are derived ~rom the motion vector f~r a motion block in tlle followlng anchor frame whose pro~ection along its motion vector to the previous anchor frame has the most overlap with that bloc]c. Thus, it is not necessary, ag on the IS system, to find matching bloclcs in subsequent and previou~ anchor ` frames so as ~o find motion vector~3 suc~ ~ mv20 and mv23 ! 0~ Fig. 4- Instead, the motion vectors for a motion block in a B2 frame are mv20 (i, j) = 2/3 mv30 (k, 1) and m23 (i, j) = -1/3 mv30 (k, l), where k and l are selected, as explained in the description of Fig. 4.
Once these B frame motion vector~ are derived, they can j be used at the encoder in the various modes discussed in connection with the IS system to determine which mode produces the least residue. It i~ then only nece~sary to transmit the mode rather than the motion vectors.

' 2~78~5~
M~TA-030 - 7 -Whereas the mo~ion vector for the later anchor frame is expressed in integral numbers of pixel spacing, the motion vector for a bloc)c in a B frame may be expressed in terms o~ Eractlons of pixel spacings, This i.s because the motion vector for a B Erame is a fraction of the motion vector for a P frame. Where two B frame~
Bl and B2 are used as in the drawings, the resolution of the shorter motion vectors for B frame motion blocks is 1/3 of a pixel spacing and that for the longer motion vector is 2/3 of a pixel spacing. Therefore, when these motion vectors are used in reconstructing a frame from pixels and interpolation i9 used, a better image can be formed. In order to obtain an independent motion vector with 1/3 pixel resolution for a motion block in the B2 frame, nine times as many locations in each of the I0 and P3 frames wouid have to be examined.

The major advantages oE deriving the motion vectors for the B frame from the P frame vectors in accordance ~ith this invention lie in the fact that in most situations less residue i9 produced than in the IS
system and no motion vectors for the B frames need to be ~ent. As in an IS sy~tem, a signal indicating mode has to be sent, but it requires significantly fewer blts than a motion vector.

: ( 2~7~
M7~TA- O 3 0 ~ 8 BRIEF DESCRIPTION OF T~IE DR7~WINGS
Fig. 1 illustrates the I, ~ and P ~rames u~ed in a IS s~s~em;

Fig. 2 illustrate~ the motion vector~ used in an IS system;

. Fig. 3 illustrates the order in which the frames are processed in an IS system;

10 Fig. 4 illustrates how motion vectors for B
frames are derived from motion vectors Eor P frames in accordance with this invention;

Fig. 5 illustrates the four blocks in a B frame that can be intercepted by the projection of a motlon block in the P frame along its motion vector to a previous anchor rame;

Fig. 6 shows the input and outputs of an area overlap motion vector calculator used in an er~odiment of this invention;

Figs. 7, 8, 9 and 10 constitute a flow chart for the operation of an area overlap motion vector calculator oE this invention u~ing t~e upper left corner of a motion block in a P frame as a poin~ of reference;

' ' 2~7~

Fig. 11 is a block ~iagram oE an encoder for a system employing independent search, IS;

F'lg. 12 i9 a bloc]c diagram of means for selecting the mode for predicting B frames and for forming the pred.icted B Erame in an IS system;

Fig. 13 is a block diagram of the decoder for an IS system;

Fig. 14 is a block diagram of an encoder using the area overlap method of this invention for determining motion vectors for predicted blocks in B frames;

Fig. 15 is a block diagram of one means for selecting the mode for predicting B frames and for forming the predicted B frames;

Fig. 16 is a bloc]c diagram of a second means for selecting the mode for predicting B frames and for forming the predicted B frames;

Fig. 17 is a block diagram of a decoder for a system using the area overlap method of this invention for determining motion vectors;

Fig, 18 illustrates the order in which frames are received and processed at the decoder; and : .
.

, - 2~7~0 Fig. 19 illustrates the order in which frames are output Erom the decoder to image reproduction mean~.

DErt'~I~ED DESCRIPTION OF T~IE INVEMTION
Reference is made to Figs. 4 and 5 for an explanation of the area~overlap method used by this invention to derive the mokion vectors for a bloc]c in a B
frame. In Fig. 4, frames Io, sl, ~2 and P3 are shown.
They are partit~oned into identical rectangular block~ of pixels that are Ll pixels.wide and L2 pixel~ hlgh and located in accordance with i and j coordinates having an origin at the upper left corners of the frames. Within each frame, pixel~ are identified by x, y coordinates.

In ~ig. 4, a motion bloclc at i - 0, ~ - 0 of a P3 frame i9 shown by the shading as being projected toward the Io frame in a direction parallel to its motion vector 2. The area of intersection of the projection with. the.Bl frame i9 shown by a rectangle 4, and the area of intersection of the projection with the ~2 frame is shown by a rectangle 5.

A method for determining the area of overlap of the area of inter~ection ~ with the various blocks of the Bl frame will now be explained by xeference ~o Figs. 4 2~78~
MAT~-030 - 11 -and 5. Although any point in a motion block of P3 could be used as a reference, the upper leEt hand corner 6 is used in th:is particular example. Its projection ls shown as intersecti~g the s1 frame at 6~ in a bloc]c at j = 0, i - 2. Observation will ~how that the area o~ intersection 4 can only overlap foux block~ in ~1~ the one inter9ected by the projection of the corner 6, which is 6', the one to the right, the one below and the one on a diagonal, i.e. those at j = 0, i = 2; j = 1, i - 2, j = 0, i = 3 and j = 1,~i = 3.

-The area in pixels of the overlap with each of the four block~ can be determined by the following formula:

AO = (L1 - ¦x - u¦) (L2 - ¦~ - v¦) ~ L1 ~2 + ¦x - u¦ ¦Y - v¦ - L2 ¦x u¦ - L1 ¦Y - v¦

where L1 equals the width of a mo~ion block in pixels, L2 the height of a motion block in pixels, u, v are the values of the x, y coordinates in pixels of a reference point having a defined relationship to the projection of a motion block in the P frame onto a B Erame, and x, ~

~ , :"

2~7~
M~TA-030 - 12 -are the coordinate~ in pixels of a xeference point having the same defined relation3hip to a B frame motion block.

By way oE example, ~uppose we wish to find the area oE overlap of the projection of the motion bloclc in the upper leEt corner of the P3 frame of FicJ. 4 with the motion bloclc at j = 1, i = 2 of the Bl frame. As shown in Fig. 5, L~ = 16 pixels and ~2 = 8 pixels. ~y observation, the width of the overlap is 12 pixels and its height is 6, so that the de3ired area is.l2 x 6 = 72 pixels. If the defined relationship is that the reference point is ~he top le~t corner of a block, then u = 12 and v ~ 18 and x = 16 and ~ = 16 so that the area of overlap in accordance with the formula is:

AO = [16 - ¦16 - 12¦] [8 - ¦16 - 18¦]

= [16 -4] [~ - 2]

= 12 x 6 = 72 As illustrated by the flow charts of Figs. 6 through 10 for calculating the areas of overlap ~0, the area~ of overlap in each of the blocks of a s frame are derived for each motion block in a P frame and the vector for the block in the B fr~ne is derived from the vector 2~78~
M~TA-030 - 13 -of the motion block in the P frcime whose pro;ection has the most overlap.

ReEerence is now rnade to Figs. 6 through 10 for a description of a flow chart of a way for determining the motion vectors of a B frame in accordance with the algorithm just described. It is thought that the comments in the various bloclss malce the chart self-explanatory, but its general operation i~ as follows.
The following terms are.used:
, T - frame number;

N - (N-l) B frames between ahchor frames;

Axea Table - current best area overlap for each B
frame motion block;

Best mv - current be~t P frame motion vector ~ 15 for each B frame motion block;

; d - distance to the previous anchor frame;

(i, ~) - location of P and B frame motion blocks. From top left corner, i motion blocks down, j motion blocks ' 8 ~ ~ ~
MATA-030 - 1~ -to the right; are used as block index in loops tlBl, ~Bl), (lB2, ~B2), ~iB3, ~B3), (lB4, ~B4) - location of B frame motion blocks. From top left corner iBk motion bloclc~ down, ~Bk motion blocks to the right;

(x, y) - pixel coordinates bE top left corner of B frame motion blocks;

(tlvP, tlhP) - pixel coordinates of top left corner of P-frame motion block in the P-frame;

(u, v) - pixel coordinates of top left corner of P frame motion block projected onto B frame;

k - B frame motion block loop counter;
and AO - area overlap.
~;
.
In a block 7 of Fig. ~, the motion vectors (PMv) from the next anchor Erame to the previou9 anchor ' .

9~ ~
M~T~-030 ~ 15 -frame are supplied to the calculator, and it s~pplie~, in a manner explained in connection with Figs. 7 through 10, the forward and backward motion vectors for a B frame.

l'he blocks a, lo, 12 and 14 of Fig. 7 are for initialization. In bloc)c 16, the pixel coordinate9 of the top left corner o~ a P frame motion bloclc are found, and using the P frame motion vector~, block 1~ locates the coordinate~ of ~he top left corner of the pro~ection block onto the s frame of interest. In bloc]c 20, the block in the s frame through which ~he projection of the upper left corner of the motion block in the P frame passes is identified, and in block 22 the other three blocks where there may be overlap with the projected motion block in the P frame are identified. In block 24, k i~ set to l and the next time through the loop it will be set at 2.

In block 26, the x and y coordinates in pixels are determined for each of the blocks identified in blocks 20, 22, and 2~, and the formula for AO is applied to each of them. A~ the AO for each of the four blocks in a B frame i~ calculated, a decision is made, block 30, a~ to whether the ~O is grea~er than an AO previou~ly calculated. If it i~ larger, a block 32 updates a table for that s block, and the procedure goe~ to a block 34 which add~ 1 to k. If block 30 indicates that the AO i~
not larger than what ig in the table, the table i.~ not :' :

2~78~0 ~r~ 030 - 16 -changed, and the p~ocedure goes ~o the block 3~. A block 36 asks whether k i9 greater than 4. If not, the~ the procedure loops back to the block 26 as indicated by a line 37.

I~ Ic is greater than 4, it means that all four blocks which a projected motion block of a P frame might intersect have been processed. ~ block 3~ then iterates i and j so as go to another motion bloc]c in the P frame.
A block 40 then checks to see if all P rame motion blocks have heen processed. If not, the procedure loops back to the block 16 as indicated by the line 41. If so, i and j are set to O in block 42 and the procedure i8 set to derive the forward and backward B frame motion vectors from the selected P frame motion vector.

`~15 ! Then, for each B frame motion block, the ; vectors from the table are obtained that are for the motion block in a P frame having a pro;ection on that ~
; frame block with the most overlap, block 44. The vectors are then scaled so as to take into account the position of the B frame between two anchor frames. If Bl i9 the bloc]c under consideration, the backward vector would be -2/3 mv30 and the orward vector would be +1t3 mv30. As ;~ this is done for each B frame block, the i and j ; coordinates are iterated. ~ block 4~ as~s if all B frame blocks have been processed. If not, the procedure loops back to the block 44 as indicated by a line 45, but if .

........ ~

2~78~5 ~r~- 030 - 17 -all B frame blocks have been processed, the calculation~
are complete until it is desired to obtain the motion vectors for the block~ of another B frame.

Fig. 11 shows a block diagram of an encoder for a system which uses the independent search method previously referred to in which forward and backward vectors are derived Eor each block in a ~ frame by a block matching process. Signals from a source 51 are digitized in an A/D device 52 and applied to the I, P
~ 10 terminal o~ a switch 53. When coding an I frame, the : switch 53 i9 in contact with itY I, P terminal 90 that : the signals for this frame are transferred block by block- to the positive input of a subtracter 54. ~lso, the negative input ~9 connected to ground by a switch 61.
The blocks of digiti~ed data are tran~foxmed blocl~ by block into DCT coefficients by a Discrete Cosine Transform processor 55. These coefficients are quantized by a quantizer 56 so as to reduce the resolution of coefficients for higher frequencies.

The quantized coefficients are variable length coded by a variable length coder, VLC, 63, and the resulting bit stream i9 coupled to a FIF0 buffer 64. The output of the FIF0 buffer 64 is a bit stream having a constant bit rate and is transmitted to an output channel by a ~ignal conveyor 66. The fullnes~ of the FIF0 buffer 64 i~ examined by a buffer controller 65 that adjust9 the , ~ , ~
;''''~
, 2 ~ 0 M~TA-030 - la -step ~ize oE the quantizer 56 so as to increase or decrease the bit rate of the signal applied to the VLC 63 and thereby prevent it from overflowing. I'hus, an I
frame i~ intra-coded so that no reference i9 made to any S other frame The I frame is recovered, albeit with quantization error, by an inver.se quantizer 57 and inverse DCT, IDCT 58 and applied to an adder 59. ~t this time, the output of the adder 59 is connected via a switch 120 to the input of a series of frame delays 61a -61e.

Note that all digitized frames at the output ofthe A/D device 52 are applied to a series of frame delays 60a through 60e. The B frames immediately following the I frame just described are applied to the fr~me delays 60a - 60e, but the switch 53 is input connected to terminal B so that the B frames are merely held for future processing, while previous B frames are supplied to terminal B of switch 53 for processing.

The next P frame, P3, for example, is encoded by again placing the switch 53 in contact with its I, P
terminal and placing the switch 61 in contact with its P, B terminal. The switch 120 at the output of the adder 59 is connected to a terminal 121 that is connected to the input of a series of fr~me delays 62a - 62e.

2 ~

MAT~-030 - 19 -The input digitized I0 frame i9 now at the output of the frame delay 60c which i~ connected to the f input of a motion estimator 67, and the input P3 frame is applied to the g input of the motion estimator 67. In a manner well ]c~own to those s]c:illed in the art, the motion e~timator 67 derives the motion vector~ lilce mv30 ~ Fig.
2 that indica~e for each motion block in P3 where a matching bloclc may be found in I0. These motion vectors appear at the output FMv of the motion estimator 67 and are applied to the VLC 63 for transmission to a decoder and to a motlon compensator means 68.

he motion compensator 68 is coupled by a line 68' to the positive input of the adder 54 so as to receive the digitized input data for P3, and a switch 73 : 15 connects an input of the compensator 68 to the output of the delay 61c so as to receive the frame IO that was recovered by the inverse ~uantizer 57 and the inverse DCT
59. By using the motion vectors from the motion ~: estimator 67, the motion compensator 68 forms a predicted . 20 frame for P3 which is supplied to the adder 59 and via a line B and the switch 61 to the negative input of the subtracter 54 wherein it is subtracted from the input data for P3 by the subtracter 54 so as to derive the difference or residue signal.

This residue signal is transformed into DCT
coefficients that are quantized in a ~imilar way as the ,~

- .
'~ ' .

~7 ~*~r~- 030 20 -Io frame men~ioned earlier. ~lso, the operatlons performed by the VLC 63, the FIFO 64, the inverse quantizer 57 and the inverse DCT 58 are the same as when the I ~rame was processed. The output of the IDCT 58 represents residue and i9 added ~o the predlcted P3 fxame by the adder 59, and the adder output is stored in a series of frame dela~s 62a - 62e. Note that the processed anchor frames are alternately stored in the frame delays 61a - 61e and 62a - 62e. Thus, I0 was applied to the delays 61a - 61e.

At this point, the encoder has in memory the original data for the I0 frame and a P~ frame that ha~
been derived from the I0 frame in the manner described.
At a decoder, the coded I0 frame iis decoded and a predicted P frame is derived from it with the aid of the motion vectors like mv30. The coded residue signal is decoded and added to the predicted P frame to produce the P3 frame In an IS system, the best mode for forming a frame from the I and P frame on either side of it is determined iIl the following way. ~he input Bl frame i9 now available at the output of the ~rame delay 60c that is connected to.the f input of the motion estimator 67.
I0 is at the output of the frame delay 60d and is connected to the input h of the motion estimator by positioning a switch 77, and the input frame P3, which i9 2 ~
M~I'A-030 - 21 -at the outpu~ of the frc~me delay 60a, i9 coupled to the input e of the motion estimator 67 by positioning a switch 76. For every motion bloc]c in sl, ~he motion estimator 67 finds a matching bloclc in the previou~
anchor frame I0 so as to produce a Eorward motion vector FMv ~uch a~ mvl0 oE Fl~. 2. For every motioll bloc]c i~ H
a matching block is also found ln the next anchor frame P3. This results in a backward motion vector BMv such as mv13 in Fig. 2.

10One or both of the motion vectors derived as just described may be sent to the VLC 63 for coding based on a decision made by the motion compensator 68 in a manner illustrated in Fig. 12. The selection made i9 called a mode, and a signal indicating the mode is sent ` 15via a line M to the VLC 63 for encoding. The B frames are not trahsmitted, but they are reconstructed at a decoder from one or both of the anchor frames by using the motion vectors in accordance with the mode signal.
Thus, if the mode signal indicates that the matching block in P3 gives best results, that matching block is used. If the matching block in I0 i9 best, it i9 . used, and if a combination of blocks produces the best results, the combination i9 used. This in done on a block by block basis so that different modes may apply to differen~ blocks.

.

2~78~0 M~r~-030 - 22 -In addition to motion vectors and mode signals, the encoder must encode the difference between a selected matching block or combination of matching block~ and the correspondin~ bloclc in the ~ Erame as it is supplied by the A/D device 52. Thi.s i9 done by supplying the selected matc}ling bloc~ or combination of matching blocks to the negative input oE the subtracter 5~ via the line B
and the switch 61. ~he differences or residue at the output of the ~ubtracter ~ are then processed by the DCT
55, the quantizer 56, the V~C 63, the F~FO 64 and applied to the signal conveyor 66. There is no need to form a predicted block with the inverse quantizer 57 and the inverse DCT 58 as decoded s frames are not needed by the encoder and therefore need not be formed at the encoder.
Therefore, the switch 120 is set on an open terminal.

The manner in which the best mode is selected in an IS system is as follows. Whereas the motion vectors FMv and BMv for a ~ frame are derived from the actual anchor frames on either side of it, these actual frames are not available at the decoder so that the best mode must be selected on the basis of the anchor frames available at the encoder. ~nchor frames substantially identical to those recovered at a decoder were produced at the output of adder 59 and alternatively applied by the switch 120 to the series of frame delays 61a - 61e and the frame delays 62a - 62e as previously described.
Future anchor frames are supplied to an input a of the 2~7~9~
Ml~'r~- 030 ~ Z3 motion Compensator 68 by me~ns of a switch 69 and switches 71 and 72, and past anchor frames by s~itches 70, 74 and 75 to an input b of the motion compensator 68.
By way of example, when the bes~ mode is to be determined for B1, the switch 71 is placed in contact with the output oE the frame delay 62a, and the ~witch 69 i~
placed in contact with the switch 71 so that the next anchor frame P3, which is at the output of 62a is applied to the a input of the motion compensator 68; the switch 74 is placed in contact with the output oE the frame delay 61d, and the switch 70 is placed in contact with the swit~h 74 90 that the previous anchor frame I0, which is at the output of the frame dela~ 61d is applied to the b input. Also, switches 77 and 76 are appropriately set so as to choose input frames I0 and P3 respectively.

Referring to Fig. 12, the FMv and aMv motion vectors supplied by the motion estimator 67 are used to select the appropriate forward and bac]cward predicted motion blocks from the past and future anchor frames by selection blocks 80 and 89 respectively. These motion blocks are respectively supplied to Mean Square Error (MSE) calculators 85 and 84. The other input to each of these calculators is the input B frame motion block.
Additionally, a third motion block called an "interpolative predicted motion block" may be generated by performing a weighted average on the forward and backward predicted motion bloc]cs. ~or example, an . ' : . , . :
: ~ ~' ' ' ' ' ' . .

.

2~3~
MATA-030 - 2~ -amplifier 81 may multiply the output o~ the block from the past anchor frame by K and the one from t}le future anchor frame by l-K. In the case of sl, K = 2/3. The interpolative predicted motion block i~ supplied to a - 5 third MSE calculator ~tage 86.
` .
Each of these MSE calculators calculates the mean square error between the input motion bloc]c and the predicted or matching motion block. These errors are supplied to a selector 87 which determines the minimum of the three errors and con~rols a switch 88 so as to choose the predicted motion block which resulted ln the least error. This chosen predicted motion block appears on the lead ~ of the motion compensator 68 of Fig. 11 and is supplied through the switch 61 to subtracter 54. Al90, mode information as to which predicted block is selected is supplied by the selector 87 and by a lead M to the VLC
63 for coding and transmission to the receiver.
`` Specifically, the mode information tells the receiver which of the three predic~ion modes was used i~ the prediction of each motion block when frame Bl was coded.
!, Figure 13 shows the block diagram of the decoder for the independent search method IS. ~n input means 97 receives the bit rate reduced signal and supplies it to a variable length decoder (VLD) 98. The VLD separates the block data into essentially four types of information: video da~a in the form o~ quantized DCT

r :

~ .

2 ~ 0 coe~ficien~s, ~orward mo~ion vectors FMv, bac)cward motion vectors sMv~ and mode information. The quantized DCT
coefficients are dequantized b~ inverse quantizer 99 and supplied to a IDCT 100 which transEorms them into residue pixels to be added to the prediction information which will be generated in the following manner. Mote that the prediction information is absent for intra-Erames (zero pixels). The prediction for P frames is obtained a3 follows. One of the two frame stores 102 and 103 contains the previous anchor frame which is to be used in generating the prediction. U3ing the forward motion vector information, the appropriate motion block is chosen from ~he previous anchor frame. After passing through switch 107, it is supplied to an adder 101 where it is added to the residue signal to form the P frame at a terminal iog. ~ust as in the encoder, a switch 106 is operated so as to store received anchor frames alternately in frame stores I and II, but received B
frames are not stored in either frame store.

When B frames are to be predicted, the mode --information instructs the displacement blocks 104 and 105 as well as switch 107 as to the t~pe of prediction to be used (i.e. forward, backward, or interpolative). Using this information as well as the appropriate motion vectors that go along with the mode information, the proper motion block or blocks are chosen from one or both frame stores 102 and 103. Bither the forward or backward 't~r '~ 'J,~ , S ;.,~, ` ~7~0 M~TA-030 - 26 -or interpolative predicted motion block i9 supplied through the s~/itch 107 to the adder 101. ~l adder lOB
performs the weighted average of the forward and backward motion bloclcs as required Eor interpolative predlction.
Frame stores I and II each have taps for a delay of 1, 2, 3, 4 or 5 Erames in a manner ~imilar to frame delays 61 and 62 in the encoder. The prediction is passed to the adder 101 where it is added to the residue signal to form the reconstructed B frames at the terminal 109.

The reconstructed frames will appear at terminal 109 in decoder processing order, Fig. lB.
Switch 130 puts the frames back into display order as shown in Fig. 19. When a reconstructed B frarne occurs at terminal lO9 switch 130 is set so as to supply it immediately to video output terminal 131. When reconstructed anchor frame occurs at texminal 109, switch 106 i9 set so as to route it to one of the frame stores;
and switch 130 i9 set as to supply the anchor frame in the other frame store to the video output terminal 131.

For example, if frame I0 is in frame store I, and P3 is in frame store II, then Bl is the next frame to be processed as shown in Fig. 18. It is immediately sent out. Similarly, B2 is imrnediately se~t out. The next fxame to be processed is P6. It is stored in frame store I while P3 in frarne store II is sent out.

- 2~78~

, Area Overla~ Method (AO) ' I'he ~rea Overlap method o~ this invention diEfers from the Independent Search method mainly in the way that the predic~ion of s Erames i9 done. The AO
,; method uses P Erame motion vectors such as mv30 for predicting B frame motion blocks in the following way. ' In this discussion, re~erence is made to the use of motion vectors for frame P3 (with respect to I0) for the purpose of predicting B1 and B2.

Fig. 14 i9 a block diagram ~or an encoder incorporating this invention. Those components corresponding to components in F'ig. 11 are designated in the same way. Forward motion vectors for P3, such as mv30, are stored in a memory 89 for a period (delay) o-E
one or two frames as required by frames Bl and B2 respectively (see Fig. 3). The stored motion vectors are supplied ~y the Fmv Memory 89 to an Area Overlap Motion Vector Calculator 90 where forward and baclcward motion vectors AO F~v'and ~O B~v are calculated for B frame motion blocks in a manner described by reference to Figs.

4 through 10. As previously explained, a backward motion vector for a B1 bloc]c = -2/3 of motion vector mv30, and the forward motion vector is 1/3 mv30. These motion vectors are supplied to the motion compensator 68 vla a lead M'.

:

,~' ~ 2~78~0 MAT~-030 - 2~ -Motion compensation m~y be done in a number o~
ways. One way is shown in E~ig. 15; another way i9 shown in Fig. 16. Corresponding components in these figures and Fig. 12 are identi~ied in the same wa~. In Fig. 15 the AO motion vectors are supplied to displacement blocks 91 and 92 that are capable of doing spatial interpolation at subpixel locations. A weighted average is performed on the outputs of the selected displaced motion blocks by multipliers 93 and 94 and an adder 95 to yield an AO
predicted motion block at the output oE the adder. This becomes a fourth candidate to be chosen by the swit~h 8~.
The calbulations o~ mean square error are made in an MSE
calculator 96, and the results are supplied to minimum selector a7 in a manner previously explained in connection with Fig. 12. We now have four possible predicted motion blocks, and once again the selector 87 instructs switch 88 to select the predicted motion block which yields the least error. The information as to which of these modes yields the lea~t error is also sent to the receiver as before. Fig. 16 is an alternative method for motion compensation. It difEers ~rom Fig. 15 only in that the AO predicted motion block has replaced the interpolative predicted motion block as a candidate, whereas in Figure 16 the ~O predicted motion block was added to the list of candidates to be checked.

Whereas the AO motion vectors, ~O FMv and AO
BMv, derived in accordance with this invention have been 2~7~9~
MATA-030 ~9 ~, , shown as being used in a bidirectional mode~ they may also be used in unidlrectional modes in a manner similar to the FMv and BMv vectors of the IS system.

A décoder for the ~O method is shown in Fi~ure 17 in which components corresponding to those of Fig. 13 are designated in the same way. In view oE the fact that motion vectors for B frames can also be derived in accordance with this invention from motion vcctors for an anchor frame ~hat is xeferenced to a previous anchor frame, bloclcs 110 and 111 are used by the decoder for this purpose. These are identical to blocks 89 and 9~ in Fig. 14. The mode information instructs the displacement ~- blocks 104 and 105 as well as switch 107 as to the ~` prediction mode to be used for B frame motion blocks.
Otherwise, the operation of this decoder is identical to that of the decoder for the IS method.

REFERENCES:
l] S. M. E. Group, MPEG Video Simula~ion Model Three lSM3).
International Organization for Standardization, July 1990.
[2] D. S. R. Center, NBC, N. ~. Philips, and T. C. Electronics, Advanced Digital Television Sys~em Descripe:ioIl. The Advanced Television Research Consortium, Februar~ 1991.
~3] A. Puri and R. Aravind, " On comparing motion-interpolation structures for video coding,'~ i~ SPIE Visual Co~llun~ca~ions and Im~ge Proces~sing, pp. 1560-1571, 1990.

' ' , 2 ~
M~TA-030 - 30 -[4] A. Puri, R. ~ravind, B.G. Haskell and R. Leonardi, " Video Codlng with Motion-Compen~ated Interpolation for CD-ROM
~pplications," in Signal Processillg: Image Conununication, pp. 127-1~ ugust 1990.
~ ..

[5] ~. Naga~a, I. lnove, ~. Tanaka and N. l'akeguchi, "Moving Picture Coding System for Digital Storage Media Using Hybrid Coding," in Signal Processing: Image Conununication, pp. 109-116, ~ugust 1990.

.. ,_~,, . , ..... ,., . . ., ... ,~ , ...................... . . ............ .

:

Claims (9)

1. In a differential pulse code modulation transmission system utilizing bidirectional motion estimation in the transmission of frames of video information, apparatus for deriving bidirectional frame motion vectors for B frames between anchor frames comprising:

means for encoding anchor frames including intraframes I and predicted frames P, said anchor frames having a plurality of B frames between them;

means for partitioning the anchor frames and B
frames into a plurality of identical blocks;

means for providing motion vectors for each block in an anchor frame that identify the position of a matching block in a previous anchor frame;

means for defining the area of intersection of the projection of each motion block in an anchor frame along its motion vector with a B frame; and means for deriving motion vectors for a block in a B frame from the motion vector of a block in an anchor frame whose projection overlaps it the most.
2. In a differential pulse code modulation transmission system utilizing bidirectional motion estimation in the transmission of frames of video information, apparatus for deriving bidirectional frame motion vectors for B frames between anchor frames comprising:

means for encoding anchor frames including intraframes I and predicted frames P, said anchor frames having a plurality of B frames between them;

means for partitioning the anchor frames and frames into a plurality of identical blocks;

means for providing motion vectors for each block in an anchor frame that identify the position of a matching block in a previous anchor frame;

means for defining the area of intersection of the projection of each motion block in an anchor frame along its motion vector with a B frames;

means for determining for each block in a B
frame the block in an anchor frame whose area of intersection overlaps it the most; and means for deriving motion vectors for each block in a s frame from the motion vector for the last mentioned block in an anchor frame.
3. Apparatus as set forth in claim 2 wherein said means for defining said area of intersection comprises:

identifying a reference point in a block in said anchor frame having a given relative position in said block;

locating a first point having the same position in said B frame as said reference point has in said anchor frame;

locating a second point in said B frame that is displaced from said first point by a fraction of the motion vector for said block in said anchor frame, the numerator of said fraction being the number of frame intervals between said B frame and said anchor frame and the denominator of said fraction being the number of frames intervals between adjacent anchor frames; and the area of intersection with said B frame having the same relationship to said second point as the block in said anchor frame has to said first point.
4. Apparatus as set forth in claim 3 wherein said reference point is at a corner of said block in said anchor frame.
5. Apparatus as set forth in claim 3 wherein an area of intersection of the projection of a block in said anchor frame has an overlap, AO, with a block in said B frame that is defined by the following L1L2 + ¦x - u¦ ¦y - v¦ - L2 ¦x - u¦ - L1 ¦y - v¦
wherein L1 is the dimension of a block as measured along a first direction from said given corner that is parallel to one edge of a frame;

L2 is the dimension of a block as measured along a second direction from said given corner that is orthogonal to said first direction;

x and y are the coordinates along said first and second directions respectively of a point having a given relationship with respect to the block in the B
frame; and u, v are the coordinates along said first and second directions respectively of a point having said given relationship with respect to the intersection of the projection of said block in said anchor frame.
6. An encoder for a differential pulse code modulation system comprising:

means for deriving motion vectors for motion blocks of an anchor frame that are referenced to a previous anchor frame;

means for deriving first sets of forward and backward motion vectors for motion blocks of B frames that are respectively referenced to the next and previous anchor frames, the first sets of motion vectors being derived by a block matching process;

means for deriving second sets of forward and backward motion vectors for blocks in B frames that are respectively referenced to the next and previous anchor frames, the second sets of motion vectors being derived from the motion vectors for the motion blocks of an anchor frame that are referenced to the previous anchor frame;

means for making respective comparisons of blocks in anchor frames that are indicated for each motion block of a B frame by the forward motion vectors of the first sets, the backward motion vectors of the first sets and the averages of the blocks so indicated with the input blocks of the B frame so as to derive representations of the respective differences;

means for comparing the average of the blocks in anchor frames that are indicated for each motion block of B frame by the forward and backward motion vectors of the second sets so as to derive representation of the respective differences; and means for deriving a mode signal for each motion block of a B frame indicating the set of motion vectors containing the motion vector associated with the smallest difference.
7. An encoder as set forth in claim 6, wherein said means for deriving second sets of forward and backward motion vectors for blocks in B frames that are respectively referenced to the next and previous anchor frames comprises:

means for partitioning the anchor frames and B
frames into a plurality of identical blocks;

means for defining the area of intersection of the projection of each motion block in an anchor frame along its motion vector with a B frames;

means for determining for each block in a B
frame the block in an anchor frame whole area of intersection overlaps it the most; and means for deriving motion vectors for each block in a B frame from the motion vector for the last mentioned block in an anchor frame.
8. A decoder for a differential pulse code modulation system comprising:

means for deriving successive anchor frames from received encoded signals;

means for storing said anchor frames;

means for receiving first sets of forward and backward motion vectors for the motion blocks of B
frames;

means for receiving motion vectors for motion blocks in an anchor frame that are referenced to the previous anchor frame;

means for deriving from said last mentioned motion vectors second sets of forward and backward motion vectors for the motion blocks of a B frame; and means for selecting the blocks of a predicted frame from the blocks of anchor frames in accordance with motion vectors from a set of motion vectors that is indicated by a mode signal.
9. A decoder as set forth in claim 8 wherein said means for deriving said second sets of forward and backward motion vectors for the motion blocks of a B
frame comprises:

means for partitioning the anchor frames and B
frames into a plurality of identical blocks;

means for defining the area of intersection of the projection of each motion block in an anchor frame along its motion vector with a B frames;

means for determining for each block in a B
frame the block in an anchor frame whose area of intersection overlaps it the most; and means for deriving motion vectors for each block in a B frame from the motion vector for the last mentioned block in an anchor frame.
CA 2078950 1991-09-23 1992-09-23 Derivation and use of motion vectors in a differential pulse code modulation system Abandoned CA2078950A1 (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US07/764,187 1991-09-23
US07764187 US5198901A (en) 1991-09-23 1991-09-23 Derivation and use of motion vectors in a differential pulse code modulation system

Publications (1)

Publication Number Publication Date
CA2078950A1 true true CA2078950A1 (en) 1993-03-24

Family

ID=25069934

Family Applications (1)

Application Number Title Priority Date Filing Date
CA 2078950 Abandoned CA2078950A1 (en) 1991-09-23 1992-09-23 Derivation and use of motion vectors in a differential pulse code modulation system

Country Status (6)

Country Link
US (1) US5198901A (en)
JP (1) JP2740716B2 (en)
KR (1) KR100270247B1 (en)
CA (1) CA2078950A1 (en)
DE (2) DE69224250D1 (en)
EP (1) EP0534350B1 (en)

Families Citing this family (74)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5986713A (en) * 1991-05-31 1999-11-16 Kabushiki Kaisha Toshiba Video coding apparatus using inter-field prediction
US6005627A (en) * 1991-05-31 1999-12-21 Kabushiki Kaisha Toshiba Video coding apparatus
US5317397A (en) * 1991-05-31 1994-05-31 Kabushiki Kaisha Toshiba Predictive coding using spatial-temporal filtering and plural motion vectors
US5640208A (en) * 1991-06-27 1997-06-17 Sony Corporation Video signal encoding in accordance with stored parameters
JPH0595540A (en) * 1991-09-30 1993-04-16 Sony Corp Dynamic picture encoder
JPH05236466A (en) * 1992-02-25 1993-09-10 Nec Corp Device and method for inter-frame predictive image encoding for motion compensation
FR2688369B1 (en) * 1992-03-03 1996-02-09 Thomson Csf Image coding method very low bandwidth and device codec óoeuvre using this method.
JPH0620050A (en) * 1992-07-03 1994-01-28 Matsushita Electric Ind Co Ltd Dynamic image signal decoding device and estimated motion vector calculation method
DE69418372T2 (en) * 1993-03-15 1999-11-25 Sony Corp Transmission and decoding of image signals
US5329318A (en) * 1993-05-13 1994-07-12 Intel Corporation Method for optimizing image motion estimation
US5684534A (en) * 1993-05-26 1997-11-04 Intel Corporation Task-splitting dual-processor system for motion estimation processing
WO1994029999A1 (en) * 1993-06-16 1994-12-22 Gould Kim V W System and method for transmitting video material
GB9315775D0 (en) * 1993-07-30 1993-09-15 British Telecomm Processing image data
DE69426584D1 (en) * 1993-08-06 2001-02-22 Lg Electronics Inc Means for converting the frame frequency
US5471252A (en) * 1993-11-03 1995-11-28 Matsushita Electric Industrial Corporation Of America Method and apparatus for estimating motion vector fields by rejecting local outliers
JP3606597B2 (en) * 1994-01-21 2005-01-05 ソニー株式会社 Motion vector detecting device
GB9401989D0 (en) * 1994-02-02 1994-03-30 British Broadcasting Corp Video motion estimation
CA2144253C (en) * 1994-04-01 1999-09-21 Bruce F. Naylor System and method of generating compressed video graphics images
KR100287211B1 (en) * 1994-08-30 2001-04-16 윤종용 Bidirectional motion estimation method and system
US5990961A (en) * 1994-09-29 1999-11-23 Telenor As Method for compression encoding of digital picture video signals, using co-encoding of blocks from several pictures
US5550595A (en) * 1994-12-16 1996-08-27 Intel Corporation Apparatus and method for motion estimation with enhanced camera interface
US5886743A (en) * 1994-12-28 1999-03-23 Hyundai Electronics Industries Co. Ltd. Object-by information coding apparatus and method thereof for MPEG-4 picture instrument
US6020926A (en) * 1994-12-30 2000-02-01 Intel Corporation Motion estimation block matching speedup
US5644504A (en) * 1995-03-27 1997-07-01 International Business Machines Corporation Dynamically partitionable digital video encoder processor
US5691775A (en) * 1995-03-30 1997-11-25 Intel Corporation Reduction of motion estimation artifacts
US5828787A (en) * 1995-04-21 1998-10-27 Sony Corporation Encoding apparatus and encoding method of motion image
US5818969A (en) * 1995-05-12 1998-10-06 Intel Corporation Intelligent start for motion estimation search
US5703966A (en) * 1995-06-27 1997-12-30 Intel Corporation Block selection using motion estimation error
US5751861A (en) * 1995-06-30 1998-05-12 Intel Corporation Reducing residual artifacts in video coding schemes with integer motion compensation
US5886744A (en) * 1995-09-08 1999-03-23 Intel Corporation Method and apparatus for filtering jitter from motion estimation video data
KR100387229B1 (en) * 1995-11-01 2003-08-21 삼성전자주식회사 Apparatus for encoding images
US5706054A (en) * 1995-12-01 1998-01-06 Intel Corporation Method and apparatus for adjusting video data to limit the effects of automatic focusing control on motion estimation video coders
FR2742248B1 (en) * 1995-12-06 1998-01-23 Thomson Multimedia Sa Data Processing Method in matrix networks in a motion estimation system
US6075906A (en) * 1995-12-13 2000-06-13 Silicon Graphics Inc. System and method for the scaling of image streams that use motion vectors
US6957350B1 (en) 1996-01-30 2005-10-18 Dolby Laboratories Licensing Corporation Encrypted and watermarked temporal and resolution layering in advanced television
JPH09212650A (en) * 1996-02-05 1997-08-15 Sony Corp Motion vector detector and detection method
US6020924A (en) * 1996-04-12 2000-02-01 Samsung Electronics Co., Ltd. Reduced memory size set top box which stores frames and associated motion vectors which indicate which block or blocks are to be retrieved from memory
US5818533A (en) * 1996-08-08 1998-10-06 Lsi Logic Corporation Method and apparatus for decoding B frames in video codecs with minimal memory
US6191781B1 (en) 1996-08-14 2001-02-20 Samsung Electronics, Ltd. Television graphical user interface that combines electronic program guide with graphical channel changer
US6181333B1 (en) 1996-08-14 2001-01-30 Samsung Electronics Co., Ltd. Television graphical user interface having channel and program sorting capabilities
US5835156A (en) * 1996-08-14 1998-11-10 Samsung Electroncis, Ltd. Television graphical user interface employing remote random access pointing device
US5955988A (en) * 1996-08-14 1999-09-21 Samsung Electronics Co., Ltd. Graphical user interface for establishing installation location for satellite based television system
US6195089B1 (en) 1996-08-14 2001-02-27 Samsung Electronics Co., Ltd. Television graphical user interface having variable channel changer icons
US5978043A (en) * 1996-08-14 1999-11-02 Samsung Electronics Co., Ltd. TV graphical user interface that provides customized lists of programming
US6411308B1 (en) 1996-08-14 2002-06-25 Samsung Electronics Co., Ltd. Television graphical user interface having variable channel control bars
US6016144A (en) * 1996-08-14 2000-01-18 Samsung Electronics Co., Ltd. Multi-layered television graphical user interface
US5940072A (en) * 1996-08-15 1999-08-17 Samsung Information Systems America Graphics decompression using system ROM indexing in TV set top box
US6037933A (en) * 1996-11-13 2000-03-14 Samsung Electronics Co., Ltd. TV graphical user interface for providing user access to preset time periods of TV program information
EP0909097B1 (en) * 1997-03-19 2006-06-07 Sony Corporation Image encoding method and device
US6404813B1 (en) 1997-03-27 2002-06-11 At&T Corp. Bidirectionally predicted pictures or video object planes for efficient and flexible video coding
JP4142180B2 (en) * 1998-10-29 2008-08-27 富士通株式会社 Motion vector coding apparatus and a decoding apparatus
US6625216B1 (en) 1999-01-27 2003-09-23 Matsushita Electic Industrial Co., Ltd. Motion estimation using orthogonal transform-domain block matching
US6473460B1 (en) 2000-03-31 2002-10-29 Matsushita Electric Industrial Co., Ltd. Method and apparatus for calculating motion vectors
GB2362532B (en) * 2000-05-15 2004-05-05 Nokia Mobile Phones Ltd Video coding
KR100708091B1 (en) * 2000-06-13 2007-04-16 삼성전자주식회사 Frame rate converter using bidirectional motion vector and method thereof
US8374237B2 (en) 2001-03-02 2013-02-12 Dolby Laboratories Licensing Corporation High precision encoding and decoding of video images
US7266150B2 (en) 2001-07-11 2007-09-04 Dolby Laboratories, Inc. Interpolation of video compression frames
US6816552B2 (en) * 2001-07-11 2004-11-09 Dolby Laboratories Licensing Corporation Interpolation of video compression frames
US8111754B1 (en) 2001-07-11 2012-02-07 Dolby Laboratories Licensing Corporation Interpolation of video compression frames
US20030112863A1 (en) 2001-07-12 2003-06-19 Demos Gary A. Method and system for improving compressed image chroma information
KR100412501B1 (en) * 2001-11-30 2003-12-31 삼성전자주식회사 Pixel-data selection device for motion compensation and method of the same
EP1347650A1 (en) * 2002-03-18 2003-09-24 Stmicroelectronics, Ltd. Compression circuitry for generating an encoded bitstream from a plurality of video frames
JP4462823B2 (en) * 2002-11-20 2010-05-12 ソニー株式会社 Processing apparatus and processing method for an image signal, generating apparatus and method for generating coefficient data used therein, and a program for performing each method
JP4053490B2 (en) * 2003-03-25 2008-02-27 株式会社東芝 Interpolation image generating method for frame interpolation and the image display system using the same, the interpolation image generating apparatus
US7310371B2 (en) * 2003-05-30 2007-12-18 Lsi Corporation Method and/or apparatus for reducing the complexity of H.264 B-frame encoding using selective reconstruction
CN1225127C (en) * 2003-09-12 2005-10-26 中国科学院计算技术研究所 A coding/decoding end bothway prediction method for video coding
US7889792B2 (en) 2003-12-24 2011-02-15 Apple Inc. Method and system for video encoding using a variable number of B frames
US7492820B2 (en) 2004-02-06 2009-02-17 Apple Inc. Rate control for video coder employing adaptive linear regression bits modeling
US7986731B2 (en) 2004-02-06 2011-07-26 Apple Inc. H.264/AVC coder incorporating rate and quality controller
US7869503B2 (en) 2004-02-06 2011-01-11 Apple Inc. Rate and quality controller for H.264/AVC video coder and scene analyzer therefor
US7453938B2 (en) 2004-02-06 2008-11-18 Apple Inc. Target bitrate estimator, picture activity and buffer management in rate control for video coder
KR101196429B1 (en) * 2004-03-12 2012-11-01 삼성전자주식회사 Video transcoding method and apparatus, and motion vector interpolation method
US7456760B2 (en) 2006-09-11 2008-11-25 Apple Inc. Complexity-aware encoding
WO2010017166A8 (en) 2008-08-04 2010-09-23 Dolby Laboratories Licensing Corporation Overlapped block disparity estimation and compensation architecture

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4985768A (en) * 1989-01-20 1991-01-15 Victor Company Of Japan, Ltd. Inter-frame predictive encoding system with encoded and transmitted prediction error
JPH0832047B2 (en) * 1989-04-28 1996-03-27 日本ビクター株式会社 Predictive coding apparatus
US5028996A (en) * 1989-06-26 1991-07-02 Matsushita Electric Industrial Co., Ltd. Picture coding method
JP3159309B2 (en) * 1989-09-27 2001-04-23 ソニー株式会社 Video signal coding method and a video signal encoding apparatus
JPH07112284B2 (en) * 1990-01-20 1995-11-29 日本ビクター株式会社 Predictive coding apparatus and decoding apparatus
US4999705A (en) * 1990-05-03 1991-03-12 At&T Bell Laboratories Three dimensional motion compensated video coding

Also Published As

Publication number Publication date Type
EP0534350A2 (en) 1993-03-31 application
JPH06217293A (en) 1994-08-05 application
DE69224250D1 (en) 1998-03-05 grant
KR100270247B1 (en) 2000-10-16 grant
EP0534350B1 (en) 1998-01-28 grant
DE69224250T2 (en) 1998-09-03 grant
JP2740716B2 (en) 1998-04-15 grant
EP0534350A3 (en) 1994-07-20 application
US5198901A (en) 1993-03-30 grant

Similar Documents

Publication Publication Date Title
US5612743A (en) Method for encoding a video signal using feature point based motion estimation
US5598216A (en) Method and apparatus for encoding/decoding a video signal
US6591015B1 (en) Video coding method and apparatus with motion compensation and motion vector estimator
US6360017B1 (en) Perceptual-based spatio-temporal segmentation for motion estimation
US5859668A (en) Prediction mode selecting device in moving image coder
US6229849B1 (en) Coding device and method
US5398068A (en) Method and apparatus for determining motion vectors for image sequences
US6205178B1 (en) Method and synthesizing a predicted image, video coding device and video coding method
US6757330B1 (en) Efficient implementation of half-pixel motion prediction
US6282243B1 (en) Apparatus and method for interframe predictive video coding and decoding with capabilities to avoid rounding error accumulation
US6122318A (en) Video encoding apparatus and video decoding apparatus
US5210605A (en) Method and apparatus for determining motion vectors for image sequences
US5805221A (en) Video signal coding system employing segmentation technique
US20080240247A1 (en) Method of encoding and decoding motion model parameters and video encoding and decoding method and apparatus using motion model parameters
US5155594A (en) Hierarchical encoding method and apparatus employing background references for efficiently communicating image sequences
US20050002458A1 (en) Spatial scalable compression
US5293229A (en) Apparatus and method for processing groups of fields in a video data compression system
US7486734B2 (en) Decoding and coding method of moving image signal, and decoding and coding apparatus of moving image signal using the same
US4703350A (en) Method and apparatus for efficiently communicating image sequences
US4805017A (en) System for coding and transmitting motion image signals
US6825885B2 (en) Motion information coding and decoding method
US6430223B1 (en) Motion prediction apparatus and method
US6754270B1 (en) Encoding high-definition video using overlapping panels
US5886742A (en) Video coding device and video decoding device with a motion compensated interframe prediction
US5689306A (en) Method and apparatus for encoding a video signal using pixel-by-pixel motion prediction

Legal Events

Date Code Title Description
EEER Examination request
FZDE Dead