GB2348559A - Motion vector estimation using multi-resolution techniques - Google Patents
Motion vector estimation using multi-resolution techniques Download PDFInfo
- Publication number
- GB2348559A GB2348559A GB9907327A GB9907327A GB2348559A GB 2348559 A GB2348559 A GB 2348559A GB 9907327 A GB9907327 A GB 9907327A GB 9907327 A GB9907327 A GB 9907327A GB 2348559 A GB2348559 A GB 2348559A
- Authority
- GB
- United Kingdom
- Prior art keywords
- layer
- mad
- search
- resolution
- minimized
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/53—Multi-resolution motion estimation; Hierarchical motion estimation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/56—Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
Abstract
A motion estimating method comprising the steps of: <SL> <LI>(a) reducing the resolution of an input frame(layer 0) to produce a layer 1 image <LI>(b) reducing the resolution of layer 1 to produce a layer 2 image <LI>(c) calculating a mean absolute difference (MAD) with respect to a search region of the layer 2, determining two positions in which the MAD calculated is minimized and using these two points as initial search centre points in layer 1 <LI>(d) determining one addition search centre point in layer 1 using a correlation of neighbourhood blocks of a current block, <LI>(e) calculating a MAD with respect to search regions using the initial search centre points in the layer 1 <LI>(f) selecting a position in which the MAD calculated in the step (e) is minimized as the search centre point in the layer O, <LI>(g) calculating a MAD with respect to a search region using the initial search centre points selected in the step (f) as a centre in the layer 0, and <LI>(h) determining a final motion vector from information on the distance between the position in which the MAD calculated in the step (g) and the origin. </SL>
Description
HIGH SPEED MOTION ESTIMATING METHOD FOR REAL TIME MOVING
IMAGE CODING AND APPARATS THEREFOR
The present invention relates to a superhigh speed motion estimating method for real time moving image coding, and more particularly, to a motion estimating method for reducing calculation of motion vector by determining a plurality of motion vector candidates under the lowered resolution in motion estimation for motion picture coding such as MPEG-2, determining motion vector candidates using motion vector correlation of neighbourhood blocks, selecting search area with a plurality of motion vector as centres, and calculating a motion vector.
Motion compensation coding for removing a temporal duplication in order to obtain a high data compression rate is widely used in moving image coding and takes an important role in International Video Coding Standard such as MPEG-1,2, and 4 or H-263 Standard.
Motion compensation coding predicts an image which is the most similar to an image received from information of a previous frame through motion estimation and conversion codes a subtraction image obtained by subtracting an estimated image from a received image.
A general apparatus for moving image coding is shown in Figure 1. Referring to Figure 1, a general moving image coding apparatus includes a frame memory 102, motion estimators 104 and 106, a motion compensator 108, a subtracter 110, a discrete cosine transformer 112, a quantizer 114, an inverse discrete cosine transformer 118, an adder 120, a frame delay 122, a forward analysis and coding rate controller 124, a variable length encoder 126, and a buffer 128.
An image input in units of a frame is stored in the frame memory 102. The first motion estimator 104 calculates a motion vector with respect to an image in units of an integer number of pixel. The second motion estimator 106 calculates a motion vector in units of a half-pixel with respect to a currently input image using a motion vector received from the first motion estimator 104, a video signal received from the frame memory 102, and information of a previous frame received from the frame delay 122.
The motion compensator 108 performs motion compensation based on the motion vector received from the second motion estimator 106 and the information of the previous frame received from the frame delay 122, so that outputs an estimated image with respect to a current frame. The temporal duplication of the moving image is removed by obtaining the subtraction image obtained by subtracting the motion compensated estimated image from the current image received from the frame memory 102 by the subtracter 110. Such motion estimating and compensating processes are performed in units of a 16X16 block, in general, such a block is called a macro block.
The subtraction image obtained after the motion estimation and compensation is discrete cosine transformed and quantized by the discrete cosine transformer 112 and the quantizer 114. Accordingly, a remaining spacial duplication of the subtraction image is removed. The motion vectors and the quantized subtraction image are encoded by the variable length encoder 126 and are transferred in a pattern of a bit stream through the buffer 128.
The forward analysis and coding rate controller 124 controls the coding rate of the variable length coder 126.
The quantized video signal is interpolated and restored by the inverse quantizer 116 and the inverse discrete cosine transformer 118. The interpolation and restoration result is added to the motion compensated video signal by the adder 120 and is stored in the frame delay 122. The image frame stored in the frame delayer 122 corresponds to an image frame the immediately preceding frame. The previous image frame information stored in the frame delayer 122 is input to the second motion estimator 106 and the motion compensator 108.
Currently, a method for estimating and compensating a motion in units of a frame and a method for estimating and compensating a motion in units of a field in a method for estimating and compensating a motion for processing a moving image are known to those skilled in the art, a description of which will be omitted in the present specification.
In a conventional method for estimating a motion by a full scale block matching analysis (FSBMA), a twodimensional motion vector of each block is estimated by dividing a current frame into blocks having a uniform size, comparing the respective blocks with all the blocks in the search region of a reference frame according to a given matching standard, and searching the position of an optimal matching block. A mean absolute difference MAD which is a relatively simple calculation is used as a matching standard for searching the optimal matching block in such a conventional block matching method. The MAD is calculated using Equation 1.
MAD (i, j) =
If (k, 1)-ft~1 (k+i, 1+j) ... (1) wherein, f (k l) is the brightness value of a pixel in a (k, 1) position of the current frame, and ,-,,. I +j) is the brightness value of a pixel in a position moved by (i, j) from (k, 1).
In such a block matching method, the maximum motion estimation scope is determined considering the motion of real images when the coding is performed. The FSBMA for estimating the motion vector by comparing all the blocks in the motion estimation scope with current blocks has the highest performance considering an estimated gain.
However, an excessive amount of calculation is required.
For example, when the maximum movement displacement in a frame is p (a pulse/a frame) with respect to a block of
MxN size, the size of the search region is (M+2p) x (N+2p) in a reference frame. Namely, since the number of candidate blocks to be compared according to the matching standard is (2p+1)2, it becomes more difficult to realize real time moving image encoding as p becomes larger.
Another conventional technology for solving such a problem is provided in"A Fast Hierarchical Motion vector
Estimation Algorithm Using Mean Pyramid", K. M. Nam, J. S.
Kim, R. H. Park, Y. S. Shim, IEEE Trans. of Circuits &
Systems for Video Technology, 1995,5, (4), pp. 344-351 and"Accuracy Improvement And Cost Reduction of 3-step
Search Region Matching Algorithm for Video Coding", IEEE
Trans. Circuits & Systems for Video Technology, 1994,4, (1), pp. 88-90. In the above documents, high speed hierarchical search methods using a plurality of candidates which can replace the FSBMA are described.
The method of using a plurality of candidates can solve a problem of a local minimum value which occurs due to a hierarchical search. However, a large amount of calculation is still required in order to achieve a performance comparable to that of the FSBMA. Also, since the methods are based on a three-step hierarchical searching method, there is a disadvantage in that it is not suitable for estimating a motion in a wide search region.
With a view to solve or reduce the above problem, it is an aim of preferred embodiments of the present invention to provide a motion estimating method by which it is possible to estimate a motion vector at a high speed by reducing the amount of calculation for calculating the motion vector.
It is another aim of preferred embodiments of the present invention to provide a motion estimating apparatus realized by the motion vector estimating method.
According to a first aspect of the invention, there is provided a motion estimating method using block matching in order to compress a moving image, comprising the steps of: (a) providing a layer 1 representing a reduced resolution image frame by reducing the resolution of a layer 0 representing an image frame of nominal resolution and further providing a layer 2 by reducing the resolution of the layer 1, wherein the layer 0 is an input image frame; (b) calculating a mean absolute difference (MAD) with respect to a search region of the layer 2; (c) determining at least two positions in which the MAD calculated in the step (b) is minimized as initial search centre points in the layer 1; (d) further determining at least one initial search centre point in the layer 1 using a correlation of neighbourhood blocks of a current block; (e) calculating a MAD with respect to search regions using the initial search centre points as a centre in the layer 1; (f) selecting a position in which the MAD calculated in the step (e) is minimized as the search centre point in the layer 0 ; (g) calculating a MAD with respect to a search region using the initial search centre points selected in the step (f) as a centre in the layer 0; and (h) determining a final motion vector from information on the distance between the position in which the MAD is calculated in the step (g) and the origin.
According to a second aspect of the invention, there is provided a motion estimating method using block matching in order to compress a moving image, comprising the steps of:
(a) providing a layer 1 representing a reduced resolution image frame by reducing the resolution of a layer 0 and further providing a layer 2 by reducing the resolution of the layer 1 when an input image frame is the layer 0;
(b) calculating a mean absolute difference (MAD) with respect to a search region of the layer 2;
(c) determining two positions in which the MAD calculated in the step (b) is minimized as initial search centre points in the layer 1;
(d) further determining an initial search centre point using a correlation of neighbourhood blocks of a current block;
(e) calculating a MAD with respect to a search region using three initial search centre points as a search centre in the layer 1;
(f) selecting a position in which the MAD calculated in the step (e) is minimized as an initial search centre point in the layer;
(g) calculating a MAD with respect to the search region of the layer 0; and
(h) determining a final motion vector using information on the distance between the position in which the MAD calculated in the step (g) is minimized and the origin.
Preferably, the step (a) provides layer 1 and layer 2 by reducing the resolution of input layer 0 to % h in horizontal and vertical directions.
Preferably, the step (d) selects the position in which the MAD is minimized as an initial search centre point in the layer 1 by calculating the mean value of the motion vectors of blocks in which an absolute subtraction of the motion vectors between peripheral blocks is no more than a predetermined threshold value.
A search region in the steps (e) and (g) may be comprised of 5x5 pixels.
According to a third aspect of the invention, there is provided a motion estimating apparatus using block matching in order to compress a moving image, comprising: resolution controlling means for providing a layer 1 by reducing a resolution of a layer 0 and providing a layer 2 by reducing a resolution of the layer 1, wherein the layer 0 is an input image frame; MAD calculating means for calculating a first MAD with respect to the search region of the layer 2, a second MAD with respect to a search region using positions corresponding to motion vector candidates in the layer 1 as a search centre point, and a third MAD with respect to a search region using the position in which the second MAD is minimized in the layer 0 as the search centre point of the layer 0; initial search centre point determining means for determining at least two positions in which the first MAD is minimized as initial search centre points in the layer 1 and further selecting at least one initial search centre points using a correlation of neighbourhood blocks of a current block; and final motion vector determining means for determining a final motion vector using information on the distance between the position in which the third MAD is minimized and the origin point.
The resolution controlling means may provide the layer 1 and the layer 2 by reducing the resolution of the received layer 0 to h and, respectively.
Preferably, the initial search centre point determining means calculates the mean values of the motion vectors of blocks in which an absolute subtraction of the motion vectors between neighbourhood blocks is no more than a predetermined threshold value and selects the position in which the mean value of the motion vectors is minimized as the initial search centre point in the layer 1.
The search region is preferably comprised of 5x5 pixels.
For a better understanding of the invention, and to show how embodiments of the same may be carried into effect, reference will now be made, by way of example, to the accompanying diagrammatic drawings, in which:
Figure 1 is a block diagram showing the schematic structure of a moving image coding apparatus to which the present invention is applied ;
Figure 2 is a flowchart showing the main steps of a motion estimating method according to an embodiment of the present invention;
Figure 3 describes hierarchical searching positions when a hierarchical search is performed by a motion estimating method according to the embodiment of the present invention; and
Figure 4 describes processes of determining motion vector candidates using the correlation of a current block with respect to peripheral blocks in the motion estimating method according to the embodiment of the present invention.
Hereinafter, preferred embodiments of-a superhigh speed motion estimating method for real time moving image coding according to the present invention and an apparatus therefor will be described in detail with reference to the attached drawings.
Figure 2 is a flowchart showing the main steps of a motion estimating method according to an embodiment of the present invention. Figure 3 describes hierarchical searching positions when a hierarchical search is performed by the motion estimating method. Referring to
Figure 2, in the motion estimating method according to embodiments of the present invention, the resolution of a received image frame (layer 0) is reduced to h in horizontal and vertical directions, as compared with its normal resolution, to provide a lower resolution image hereinafter referred to as layer 1, and the resolution of layer 0 is reduce to h in horizontal and vertical directions to provide a further reduced resolution image hereinafter referred to as layer 2 (step 202). The resolution is reduced in order to reduce the amount of calculation for calculating the motion vector by reducing the size of the search region. For example, when the size of the entire search region is Rx R, the range of the search region becomes (R/4) x (R/4) in the present embodiment.
A mean absolute difference (MAD) is calculated with respect to the search region (Figure 3) of the layer 2 (step 204). Since the search SR (Figure 3) is determined to be a region from N to s when when N is a predetermined positive integer, it s s becomes a region 2N in width and length including 0, 2/y thus the number of the search positions will be (2Nsx2Ns)/16 e., (lVsxlVs)/4' And then, two positions in which the MAD calculated in the step 204 is minimized, namely, thus, an estimated error is smallest, are selected as an initial search centre point of a next step (in the layer 1) (step 206).
One initial search centre point is further selected.
Processes of selecting further initial search centre point will be described with reference to Figure 4. The compression of a digital moving image is performed in units of a block such as a macro block. For example, the blocks can be sequentially processed row by row.
Peripheral blocks used in embodiments of the present invention become mutually related three blocks as shown in
Figure 4. Also, the peripheral blocks can be divided into five groups as shown in Figure 4, considering the correlation between motion vectors. For example, when the motion vector MVs of the peripheral blocks are as shown in
Figure 4A, an absolute subtraction of the motion vector between the peripheral blocks is defined by Equation 2.
#=tJM-MV,)),
(2 = ls2-F31,... (2
#3 + #MV3-MV1# And then, D is defined as a threshold value for checking the similarity between two motion vectors. Next, the peripheral blocks are divided into five groups, considering the correlation between the motion vectors.
Namely, referring to Figure 4B, group 1 satisfies the lconditions #1#D, #2#D, and #3#D. Referring to Figure 4C, group 2 satisfies the conditions #1#D, #2# > D, and #3 > D Referring to Figure 4D, group 3 satisfies the conditions #1 > D #2#D and #3 > D. Referring to Figure 4E, group 4 satisfies the conditions #1 > D, #2 > D, and #3#D. If otherwise, it corresponds to group 5. The motion vectors are respectively calculated as MV (MV1+MV2+MV3)/3' MV= (MVi+M1)/2'MV= (MV2+MV3)/2, and MV=(MV1+MV3)/2 in case of groups 1,2,3, and 4 respectively. The position where the mean value of the calculated motion vectors is minimized is selected as an initial search centre point.
In this embodiment, one method of dividing the peripheral blocks into five groups according to the absolute subtraction of the motion vector and determining the initial search point by calculating the mean motion vector in each group is described only for the sake of explanation, it can be modified into another method of selecting one initial search centre point using the motion vector correlation of the peripheral blocks, and thus it does not restrict the scope of the present invention defined by the attached claims.
Then, the MADs of the search regions SR1(1), SR2(1), and SR3(1) based on three initial search centre points are calculated in the layer 1 (step 210). In the present embodiment, the search region in a first layer is set to be formed of 2 pixels with respect to a search centre point, i. e., 5x5 pixels in total.
Then, the position where the MAD calculated in the step 210 is minimized is selected as the initial search point in the layer 0 (step 212). The MAD is calculated using 2 pixels of region with respect to the selected initial search centre point as the search region of the layer 0 (step 214).
The final motion vector is determined using information on the distance between the position in which the calculated MAD is minimized and the origin point (step 216). The determined final motion vector becomes x-and yaxis coordinate values in the case of a two-dimensional image.
A simulation experiment using a MPEG-2 experiment image was performed in order to estimate the performance of the motion estimating method according to the present invention. Five MPEG-2 image sequences, i. e.,'Football' (fb), Cart (car), Susie (sus) tMobile and CalendarZ (mob), and'Cheer-leaders' (cheer) are used as input images among the images known to those skilled in the art as experiment images.
All the images have a frame rate of 30 Hz and is comprised of 150 frames respectively. The size of each frame is 720x480. Search regions of 64x64 (Ns=32), 10x10, and 5x5 having search point distances of 4,2, and 1, respectively, were selected in the layers 2,1, and 0.
The size NB of the block is 16 in the layer 0. The threshold value D for investigating the correlation between the peripheral blocks is set to be 8 in order to consider that a similarity is lowered when it deviates from 8 pixels. This value corresponds to the value of twice of the search point distance in the layer 2.
In order to prove the effect of the motion estimating method according to the present invention, the performance of the FSBMA is compared with the performance of a general hierarchical searching method. Three motion vectors having only the minimum estimated difference are determined to be candidates in the layer 2 of the general hierarchical searching method in which a plurality of candidates are used. Meanwhile, a third candidate is estimated among the motion vectors of the neighbourhood blocks in the FSBMA. Table 1 shows results of comparing a motion estimation with a motion compensation with respect to the first 150 frames of the MPEG-2 experiment image.
[TABLE 1]
Classific-Number Average peak signal to noise ratio (PSNR) (dB) Complexity ation of (%) candidates car sus mob cheer FSBMA N/A 25.4 28.0 36.8 24.7 23.2 100.0 Method 3 25.2 27.8 36.6 24.5 23.0 1.5 according to the present invention Referring to Table 1, in the motion estimating method according to the present invention, the performance is slightly reduced about by 0.2dB compared to FSBMA, however, the complexity is only 1.5% that of the FSBMA.
If the number of operations and the number of blocks required for calculating the MAD per pixel are N and NB respectively, the complexity can be estimated by .Here,, , and 25
C=MNB (64 NS+ 4 x25+25) 64 1VS 4 x25 represent the proportional values of the calculation amounts according to the resolutions of the layer 2, the layer 1, and the layer 0, respectively. Therefore, the complexities of the motion estimating method according to the present invention are 1.5% and 0.7% that of those of the FSBMA with respect to the N of 32 and 64.
s
Also, Table 2 shows a result of applying the motion estimating method according to the present invention to an
MPEG-2 encoder which requires a relatively wide search region. In the present experiment, the number N of group of pictures (GOP), a distance between P frames, and an objective bit rate T are respectively set as 12,2, and 6Mbps. Table 2 shows the result of comparing the PSNRs of reproduced images.
[TABLE 2]
Classification Average PSNR (dB) Complexity fb car sus mob cheer (%) MPEG-2 encoder to which 34.3 38.6 42.6 30.0 32.0 100.0 FSBMA is applied MPEG-2 encoder to which the 34.3 38.4 42.6 29.9 32.0 0.9 method according to the present invention is applied Referring to Table 2, in the motion estimating method according to the present invention, the calculation amount is remarkably reduced to 0.9% that of the FSBMA and the
PSNR is almost equal to that of the FSBMA.
As mentioned above, in the motion estimating method according to the present invention, the calculation amount is small and the high speed motion can be estimated since the plurality of motion vector candidates and the correlation of the motion vector space, which are based on the minimum estimated difference between frames are effectively used.
The superhigh speed motion estimating method for the real time moving image coding according to the present invention can be realized as an apparatus. The motion estimating apparatus (not shown) includes a resolution controller, an MAD calculator, an initial search point centre determiner, and a final motion vector determiner.
The resolution controller provides the layer 1 by reducing the resolution of the layer 0 when the received image frame is the layer 0 and provides the layer 2 by reducing the resolution of the layer 1. For example, it is possible to realize an image frame using the mean of two adjacent pixels when the resolution is to be reduced to % and the mean of four adjacent pixels when the resolution is to be reduced to 1/4.
The MAD calculates a first MAD with respect to the search region of the layer 2, a second MAD with respect to the search region using the positions corresponding to the motion vector candidates in the layer 1 as the search centre point, and a third MAD with respect to the search region using the position in which the second MAD is minimized in the layer 0 as the search centre point of layer 0.
The initial search centre point determiner determines at least two positions in which the first MAD is minimized as initial search points in the next step (the layer 1).
Also, the initial search centre point determines further determines at least one initial search centre points as initial search centre points in the next step (the layer 1) using the correlation of the peripheral blocks of the current block. The final motion vector determiner determines the final motion vector using the information on the distance between the position in which the third
MAD is minimized and the origin.
Also, the respective steps of the motion estimating method can be written as a program which can be executed in a computer and can be realized in a general digital computer for operating the program from a medium used in the computer. The medium may be a magnetic recording medium such as a floppy disk or a harddisk, an optical recording medium such as a CD-ROM or a DVD, or a carrier wave such as a transmission through the Internet. Also, the functional program, the code, and the code segments can be easily inferred by any programmer in the technical field to which the present invention belongs.
The rader's attention is directed to all papers and documents which are filed concurrently with or previous to this specification in connection with this application and which are open to public inspection with this specification, and the contents of all such papers and documents are incorporated herein by reference.
All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and/or all of the steps of any method or process so disclosed, may be combined in any combination, except combinations where at least some of such features and/or steps are mutually exclusive.
Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. Thus, unless expressly stated otherwise, each feature disclosed is one example only of a generic series of equivalent or similar features.
The invention is not restricted to the details of the foregoing embodiment (s). The invention extends to any novel one, or any novel combination, of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to any novel one, or any novel combination, of the steps of any method or process so disclosed.
Claims (11)
- CLAIMS 1. A motion estimating method using block matching in order to compress a moving image, comprising the steps of: (a) providing a layer 1 representing a reduced resolution image frame by reducing the resolution of a layer 0 representing an image frame of nominal resolution and further providing a layer 2 by reducing the resolution of the layer 1, wherein the layer 0 is an input image frame; (b) calculating a mean absolute difference (MAD) with respect to a search region of the layer 2; (c) determining at least two positions in which the MAD calculated in the step (b) is minimized as initial search centre points in the layer 1; (d) further determining at least one initial search centre point in the layer 1 using a correlation of neighbourhood blocks of a current block; (e) calculating a MAD with respect to search regions using the initial search centre points as a centre in the layer 1; (f) selecting a position in which the MAD calculated in the step (e) is minimized as the search centre point in the layer 0; (g) calculating a MAD with respect to a search region using the initial search centre points selected in the step (f) as a centre in the layer 0; and (h) determining a final motion vector from information on the distance between the position in which the MAD is calculated in the step (g) and the origin.
- 2. A motion estimating method using block matching in order to compress a moving image, comprising the steps of: (a) providing a layer 1 representing a reduced resolution image frame by reducing the resolution of a layer 0 and further providing a layer 2 by reducing the resolution of the layer 1 when an input image frame is the layer 0; (b) calculating a mean absolute difference (MAD) with respect to a search region of the layer 2; (c) determining two positions in which the MAD calculated in the step (b) is minimized as initial search centre points in the layer 1; (d) further determining an initial search centre point using a correlation of neighbourhood blocks of a current block; (e) calculating a MAD with respect to a search region using three initial search centre points as a search centre in the layer 1; (f) selecting a position in which the MAD calculated in the step (e) is minimized as an initial search centre point in the layer; (g) calculating a MAD with respect to the search region of the layer 0; and (h) determining a final motion vector using information on the distance between the position in which the MAD calculated in the step (g) is minimized and the origin.
- 3. The method of claim 2, wherein the step (a) provides layer 1 and layer 2 by reducing the resolution of input layer 0 to e and k in horizontal and vertical directions.
- 4. The method of claim 2 or 3, wherein the step (d) selects the position in which the MAD is minimized as an initial search centre point in the layer 1 by calculating the mean value of the motion vectors of blocks in which an absolute subtraction of the motion vectors between peripheral blocks is no more than a predetermined threshold value.
- 5. The method of claim 2,3 or 4, wherein a search region in the steps (e) and (g) is comprised of 5x5 pixels.
- 6. A motion estimating apparatus using block matching in order to compress a moving image, comprising: resolution controlling means for providing a layer 1 by reducing a resolution of a layer 0 and providing a layer 2 by reducing a resolution of the layer 1, wherein the layer 0 is an input image frame; MAD calculating means for calculating a first MAD with respect to the search region of the layer 2, a second MAD with respect to a search region using positions corresponding to motion vector candidates in the layer 1 as a search centre point, and a third MAD with respect to a search region using the position in which the second MAD is minimized in the layer 0 as the search centre point of the layer 0; initial search centre point determining means for determining at least two positions in which the first MAD is minimized as initial search centre points in the layer 1 and further selecting at least one initial search centre points using a correlation of neighbourhood blocks of a current block; and final motion vector determining means for determining a final motion vector using information on the distance between the position in which the third MAD is minimized and the origin point.
- 7. The apparatus of claim 6, wherein the resolution controlling means provides the layer 1 and the layer 2 by reducing the resolution of the received layer 0 to h and respectively.
- 8. The apparatus of claim 6, wherein the initial search centre point determining means calculates the mean values of the motion vectors of blocks in which an absolute subtraction of the motion vectors between neighbourhood blocks is no more than a predetermined threshold value and selects the position in which the mean value of the motion vectors is minimized as the initial search centre point in the layer 1.
- 9. The apparatus of claim 2, wherein the search region is comprised of 5x5 pixels.
- 10. A motion estimating method substantially as herein described with reference to the accompanying drawings.
- 11. Motion estimating apparatus substantially as herein described with reference to the accompanying drawings.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB9907327A GB2348559B (en) | 1999-03-31 | 1999-03-31 | High speed motion estimating method for real time moving image coding and apparatus therefor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB9907327A GB2348559B (en) | 1999-03-31 | 1999-03-31 | High speed motion estimating method for real time moving image coding and apparatus therefor |
Publications (3)
Publication Number | Publication Date |
---|---|
GB9907327D0 GB9907327D0 (en) | 1999-05-26 |
GB2348559A true GB2348559A (en) | 2000-10-04 |
GB2348559B GB2348559B (en) | 2001-06-06 |
Family
ID=10850656
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
GB9907327A Expired - Fee Related GB2348559B (en) | 1999-03-31 | 1999-03-31 | High speed motion estimating method for real time moving image coding and apparatus therefor |
Country Status (1)
Country | Link |
---|---|
GB (1) | GB2348559B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005001625A2 (en) | 2003-06-09 | 2005-01-06 | Nvidia Corporation | Mpeg motion estimation based on dual start points |
EP1610561A2 (en) | 2004-06-27 | 2005-12-28 | Apple Computer, Inc. | Encoding and decoding images |
US7792194B2 (en) | 2003-04-10 | 2010-09-07 | Lefan Zhong | MPEG artifacts post-processed filtering architecture |
US7792188B2 (en) | 2004-06-27 | 2010-09-07 | Apple Inc. | Selecting encoding types and predictive modes for encoding video data |
US8111752B2 (en) | 2004-06-27 | 2012-02-07 | Apple Inc. | Encoding mode pruning during video encoding |
US8254459B2 (en) | 2002-07-14 | 2012-08-28 | Apple Inc. | Adaptive motion estimation |
US8731071B1 (en) | 2005-12-15 | 2014-05-20 | Nvidia Corporation | System for performing finite input response (FIR) filtering in motion estimation |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9330060B1 (en) | 2003-04-15 | 2016-05-03 | Nvidia Corporation | Method and device for encoding and decoding video image data |
US8724702B1 (en) | 2006-03-29 | 2014-05-13 | Nvidia Corporation | Methods and systems for motion estimation used in video coding |
US8660380B2 (en) | 2006-08-25 | 2014-02-25 | Nvidia Corporation | Method and system for performing two-dimensional transform on data value array with reduced power consumption |
US8756482B2 (en) | 2007-05-25 | 2014-06-17 | Nvidia Corporation | Efficient encoding/decoding of a sequence of data frames |
US9118927B2 (en) | 2007-06-13 | 2015-08-25 | Nvidia Corporation | Sub-pixel interpolation and its application in motion compensated encoding of a video signal |
US8873625B2 (en) | 2007-07-18 | 2014-10-28 | Nvidia Corporation | Enhanced compression in representing non-frame-edge blocks of image frames |
US8666181B2 (en) | 2008-12-10 | 2014-03-04 | Nvidia Corporation | Adaptive multiple engine image motion detection system and method |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5477272A (en) * | 1993-07-22 | 1995-12-19 | Gte Laboratories Incorporated | Variable-block size multi-resolution motion estimation scheme for pyramid coding |
-
1999
- 1999-03-31 GB GB9907327A patent/GB2348559B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5477272A (en) * | 1993-07-22 | 1995-12-19 | Gte Laboratories Incorporated | Variable-block size multi-resolution motion estimation scheme for pyramid coding |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8254459B2 (en) | 2002-07-14 | 2012-08-28 | Apple Inc. | Adaptive motion estimation |
US7792194B2 (en) | 2003-04-10 | 2010-09-07 | Lefan Zhong | MPEG artifacts post-processed filtering architecture |
EP1631884A4 (en) * | 2003-06-09 | 2007-05-09 | Nvidia Corp | Mpeg motion estimation based on dual start points |
WO2005001625A2 (en) | 2003-06-09 | 2005-01-06 | Nvidia Corporation | Mpeg motion estimation based on dual start points |
EP1631884A2 (en) * | 2003-06-09 | 2006-03-08 | Nvidia Corporation | Mpeg motion estimation based on dual start points |
EP1610561A3 (en) * | 2004-06-27 | 2010-07-21 | Apple Inc. | Encoding and decoding images |
US7792188B2 (en) | 2004-06-27 | 2010-09-07 | Apple Inc. | Selecting encoding types and predictive modes for encoding video data |
US8018994B2 (en) | 2004-06-27 | 2011-09-13 | Apple Inc. | Selecting encoding types and predictive modes for encoding video data |
US8111752B2 (en) | 2004-06-27 | 2012-02-07 | Apple Inc. | Encoding mode pruning during video encoding |
EP1610561A2 (en) | 2004-06-27 | 2005-12-28 | Apple Computer, Inc. | Encoding and decoding images |
US8472516B2 (en) | 2004-06-27 | 2013-06-25 | Apple Inc. | Selecting encoding types and predictive modes for encoding video data |
US8737479B2 (en) | 2004-06-27 | 2014-05-27 | Apple Inc. | Encoding mode pruning during video encoding |
US8731071B1 (en) | 2005-12-15 | 2014-05-20 | Nvidia Corporation | System for performing finite input response (FIR) filtering in motion estimation |
Also Published As
Publication number | Publication date |
---|---|
GB9907327D0 (en) | 1999-05-26 |
GB2348559B (en) | 2001-06-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6377623B1 (en) | High speed motion estimating method for real time moving image coding and apparatus therefor | |
CN112866705B (en) | Encoding and decoding method, device and equipment | |
KR100994768B1 (en) | Motion estimation method for encoding motion image, and recording medium storing a program to implement thereof | |
US6317460B1 (en) | Motion vector generation by temporal interpolation | |
GB2348559A (en) | Motion vector estimation using multi-resolution techniques | |
US20020168008A1 (en) | Method and apparatus for coding moving pictures | |
JP3615963B2 (en) | Fast motion estimation method and apparatus for real-time video coding | |
KR100448700B1 (en) | Frame rate conversion of cross fade interp0lation apparatous and method there of |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PCNP | Patent ceased through non-payment of renewal fee |
Effective date: 20180331 |