21 Candidate Position Subsampling Technique

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

1052 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 20, NO.

7, JULY 2010

2:1 Candidate Position Subsampling Technique for


Fast Optimal Motion Estimation
Hwal-Suk Lee, Student Member, IEEE, Jik-Han Jung, and Dong-Jo Park, Member, IEEE

Abstract—The candidate position subsampling technique lower bounds than the SAD, this method is quicker. The
(CPST) basically chooses candidates in a search window at a third method is to use a pixel-subsampling technique [5]–[7].
sampling rate. The 2:1 CPST chooses half the candidates, and
When calculating the SAD, not all pixels are considered, but
then selects one or more candidates that are considered as to be
close to the optimal motion vector before conducting a fine search. pixels on the subsampling pattern are considered. This method
The fine search is conducted by checking four neighbors of the guarantees an improvement in speed by as much as the rate
chosen candidate(s) referred to as winner(s). The CPST can be of pixel-subsampling.
combined with a fast optimal block-matching algorithm, such as The third method is usually combined with the candidate
the multilevel successive elimination algorithm (MSEA), in order
position subsampling technique (CPST) [5], [7]. The CPST
to reduce the number of computations used in rejecting the non-
best candidate. We propose a new 2:1 CPST fitted to the MSEA. consists of three steps. First, candidates for the optimal
The proposed algorithm adopts a new condition for the winner motion vector (MV) are selected by subsampling. Second,
which helps to find the best candidate efficiently. Moreover, a one or more candidates that are considered to be close to the
fast motion estimation step is used to reduce the number of optimal MV are selected from candidates that are chosen in
computations of the MSEA, and the peak signal-to-noise ratio
the first step. The candidates that are selected in the second
(PSNR) compensation step is adopted to guarantee that the PSNR
performance of the proposed algorithm is very close to that of step are called winners. Finally, neighbors of each winner are
the full search. Experimental results show that the proposed checked in the fine search step.
algorithm reduces the computational loads of the MSEA by Recently, there has been an attempt to combine the ad-
47.26% on average with only −0.027 dB PSNR degradation in justable partial distortion search (APDS) algorithm (an algo-
the worst case.
rithm that uses the second method of using lower bounds
Index Terms—Block matching, candidate position, fast full of the SAD [3]) and the CPST [8]. The first and second
search, motion estimation, subsampling, successive elimination. candidates that have the minimum SAD are selected as win-
I. Introduction ners from candidates to which the 2:1 subsampling pattern
is applied. Finally, the four neighbors of each winner are
M OTION estimation via a block-matching algorithm
(BMA) has been used in almost all the video coding
standards, including ITU-T H.263/H.264 and ISO/IEC MPEG-
checked. However, this algorithm does not perform well if the
multilevel successive elimination algorithm (MSEA) [4], one
of the most well-known algorithms using the second method,
1/2/4. Although the full-search (FS) algorithm is the optimal
is used instead of the APDS. This results mainly from the
BMA with regard to motion estimation accuracy, its computa-
inadequate specification of criteria for selecting the winners.
tional complexity limits practical usage. Hence, many BMAs
In this letter, we propose a 2:1 candidate position subsam-
have been developed to reduce the complexity. The methods
pling technique that is used with the MSEA. The proposed
used in BMAs can be divided into three types.
algorithm shows almost the same peak signal-to-noise ratio
The first method is to use only a subset of all the search
(PSNR) performance as a FS algorithm (in the worst case,
locations in the search window [1], [2]. This method reduces
−0.027 dB degradation) with only almost half the computa-
computational complexity because the sum of absolute differ-
tions of the MSEA. This performance is achieved by adopting
ence (SAD) is not computed for all possible candidates. The
a new condition for the winner, the fast motion estimation,
second method is to adopt lower bounds for the SAD [3], [4].
and the PSNR compensation steps.
Before calculating the SAD of the current candidate, lower
bound values are obtained, and the necessity of calculating II. Related Work
the SAD is determined by comparing lower bound values with
the minimum SAD obtained from the previous candidates. The APDS algorithm [3] and the MSEA [4] are both known
Given that fewer computations are required to compute the to be fast optimal block-matching algorithms, which means
that the algorithm can achieve the same PSNR performance
Manuscript received September 26, 2008; revised August 7, 2009 and as the FS algorithm, but with fewer computations. The basic
December 30, 2009. Date of publication May 27, 2010; date of current version
July 16, 2010. This paper was recommended by Associate Editor H. Yu. procedure of the APDS and the MSEA is to calculate lower
The authors are with the School of Electrical Engineering and Com- bounds (also called boundary values) from the lowest to
puter Science, Korea Advanced Institute of Science and Technology, Dae- the highest and to check whether each boundary value is
jeon 305-701, Korea (e-mail: seagall@kaist.ac.kr; honestok@gmail.com; dj-
park@ee.kaist.ac.kr). larger than the minimum SAD value that was obtained from
Digital Object Identifier 10.1109/TCSVT.2010.2051365 previously searched positions. This value will be referred to as
1051-8215/$26.00 
c 2010 IEEE
LEE et al.: 2:1 CANDIDATE POSITION SUBSAMPLING TECHNIQUE FOR FAST OPTIMAL MOTION ESTIMATION 1053

TABLE I
Algorithm 1: Pseudo code for the APDS and the MSEA
Number of Cases in Which the Optimal MV Is Missed With
(x, y) = initial position Different Conditions
Find SAD of (x, y)
SAD min = SAD of (x, y) Name Format Total Condition 1 Condition 2 Condition 3
(x, y) = next position Mobile CIF 12 586 1222 466 285
while !(last candidate position) do Foreman CIF 9650 2754 1792 536
for k=1 to number LB do Fun CIF 7106 622 233 219
Compute LB[k] Paris CIF 3093 325 116 59
if LB[k]>SAD min then Mobile CCIR601 15 362 2521 1304 800
go to (#) Football CCIR601 18 277 2309 946 1010
Flower CCIR601 16 908 2621 1756 1447
end
Tennis CCIR601 8178 5556 4680 2985
end Average 11395.0 2241.3 1411.6 917.6
SAD min = LB[number LB] Probability of Missing MV 0.197 0.124 0.081
MV = (x, y)
(#) (x, y) = next position Condition 1: Choose the one candidate with the minimum SAD. Condition 2: Choose the
end two candidates with the first and second minimum SADs. Condition 3: Choose candidates
updating the temporary minimum SAD.
· LB: Lower bound, MV: Motion vector, (x, y) : Current candidate

Fig. 1. Search patterns. (a) 2:1 candidate subsampling pattern. (b) Fine
search pattern.
Fig. 2. Sample plot for the proposed condition in 1-D space.

the temporary minimum SAD. If a certain lower bound of the III. Proposed 2:1 Candidate Position Subsampling
current position is larger than the temporary minimum SAD, Technique for the MSEA
it is deemed unnecessary to compute the SAD of the current
position. This procedure is presented in Algorithm 1. A. New Condition for Winner
The main difference between the APDS and the MSEA The purpose of a fine search is to find the optimal MV in
is the way of obtaining lower bounds. The APDS calculates the case that the optimal MV is not in the pattern yielded by
boundary values by using the partial distortion technique subsampling. Therefore, the best winner out of the candidates
which is derived from the fact that the SAD value for the selected by subsampling is the one that has the optimal MV
N × N block size is always larger than the partial sum of as a neighbor if the fine search checks four neighbors of
the absolute value for difference between pixel-values in the the winner, as depicted in Fig. 1(b). If the correlation between
current and reference blocks. By contrast, the MSEA obtains the SAD values of the optimal MV and its neighbors is high,
boundary values by using the triangle inequality as a basis: the candidate selected by subsampling that has the minimum
|a + b| ≤ |a| + |b|. SAD is likely to be the best winner. In most cases, it is
The CPST addresses the issue of which candidates are acceptable that the correlation is high, because the maximum
to be checked and the search order of selected candidates, distance between the winner and a neighbor is just one
while the APDS and the MSEA are the candidate-rejection pixel.
algorithms (CRAs) that correspond to the inside of a while- Our experiments show that when the condition of selecting a
loop in Algorithm 1. In [8], the CPST consists of three steps candidate that has the minimum SAD is used, the probability
with the APDS as the CRA. In the first step, candidates are of missing the optimal MV is 0.197, as shown in Table I,
searched using the candidate subsampling pattern, as depicted where the total number (Total) stands for the number of cases
in Fig. 1(a). In the second step, two candidates with the first that the optimal MV is not in the subsampling pattern. The
and second minimum SAD values are selected as winners. simulation was executed using the same conditions as those
Finally, the four candidates nearest neighbors to each winner in Table III.
are searched, as in Fig. 1(b). Missing the optimal MV means degradation of the PSNR
Given that, in the second step, two candidates with the first performance. To improve the PSNR performance, the condi-
and second minimum SAD values are selected, Algorithm 1 tion for selecting the winner needs to be changed. One way is
should be changed slightly. A noteworthy change is that a to choose multiple winners using the increasing order of the
certain lower bound is compared not with the first minimum SAD values as a basis. For instance, two candidates with the
SAD but with the second minimum SAD. This modification first and second minimum SAD are selected as winners in [8].
was also mentioned in [8]. The condition used in [8] reduces the probability of missing
1054 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 20, NO. 7, JULY 2010

the optimal MV to 0.124 in Table I, but it requires additional TABLE II


computational costs to choose the winners. Implemented Algorithms
In Algorithm 1, for the non-best candidate, the number of
Label STEP 1 STEP 2 STEP 3
the boundary values used to make the decision to reject the Conv CPST Unused Condition 1 Unused
current candidate depends heavily on the temporary minimum Prop CPST1 Unused Condition 3 Unused
SAD (SAD min in Algorithm 1). The lower the value of the Prop CPST2 Unused Condition 3 Used
temporary minimum SAD is set, the fewer boundary values are Proposed Used Condition 3 Used
used. When the CPST chooses two winners with the first and STEP 1: Fast motion estimation. STEP 2: Winner selection. STEP 3: PSNR
second minimum SAD, the temporary minimum value used to compensation.
make the decision to reject the current candidate is not the first
minimum SAD, but the second minimum SAD. This implies that of the expected optimal MV. The PSNR compensation
that computational load of the CRA is increased. step is performed when the SAD value of the final MV in the
Usage of the second minimum SAD instead of the first previous fine search step is larger than the highest expected
minimum SAD causes more boundary values to be calculated. SAD value of the optimal MV (SADmax ), because the PSNR
In the APDS, the difference in the SAD-computation numbers value is strongly related to the SAD value. In order to obtain
between adjacent boundary values is maximally 16, while that SADmax , the maximum SAD value among the SAD values
of the MSEA is 192 if the block size is 16 × 16. Thus, usage of the three MVs of the left, top, and top-right blocks of
of the second minimum SAD in the MSEA requires a much the current block are calculated, and this result is compared
greater amount of additional computation than the APDS. with a constant K. The larger value is selected as the highest
It is preferable that the condition for selecting the winner expected SAD value (SADmax ). K is adopted to prevent the
requires fewer computations and that the probability that any PSNR compensation step from being performed when SAD
winner selected by the condition is a neighbor of the optimal values of the three MVs are simultaneously small. In our case,
MV is higher. By trial and error, we found the most efficient we have selected K = 1792.
condition for selecting the winner. The proposed condition
C. Fast Motion Estimation
is that a candidate is chosen as a winner if it updates the
temporary minimum SAD value. Fig. 2 plots the difference The length of time it takes the CRA to complete its task
between the proposed condition and conventional conditions in is strongly related to the temporary minimum SAD value
1-D space. Here, D(x) indicates the SAD value of a candidate (SAD min in Algorithm 1). This implies that the CRA will
x. Note that the global minimum position C3 can be found perform its task quickly if the optimal MV is found quickly.
only if C2 or C4 are selected as winners, because the fine Hence, we insert a fast motion-estimation (ME) step before
search is conducted by checking left and right candidates of the candidate are subsampled. As a fast ME algorithm, we
the winner. In the first step, half of the candidates are checked adopt a simplified version of the algorithm proposed in [9].
at a 2:1 subsampling rate (C0 → C2 → · · · C12 ). For the fine The fast ME algorithm consists of two steps: 1) prediction
search step, if Condition 1 in Table I is used, C8 is chosen as of the optimal MV, and 2) unrestricted search with a small
the winner. Moreover, Condition 2 selects C8 and C10 as the diamond-shaped pattern (SDSP), as shown in Fig. 1(b). In the
winners. However, if the proposed condition (Condition 3) is first step, the MV with the minimum SAD is selected among
used, C0 , C2 and C8 are selected as the winners. Note that MVL , MVT , MVTR , MVMED , and (0, 0). Here, MVL , MVT ,
the first candidate (C0 in Fig. 2) is always considered to be and MVTR are MVs of the left, top, and top-right blocks of
the winner, because it initializes the temporal minimum SAD the current block respectively, and MVMED is the median of
value. MVL , MVT , and MVTR . In the second step, a SDSP is formed
Table I shows the superiority of the proposed condition. It by positioning the MV that was found in the previous step at
lessens the probability that the optimal MV will be missed the center of the SDSP. If the new minimum SAD point is still
to 0.081 without requiring any additional computation. The at the center of the formed SDSP, then ME stops. Otherwise,
proposed condition requires additional Boolean memory space this step is repeated. Note that the CRA is also used in the
that is equivalent to half of the size of search window, due to fast ME step to reject the non-best candidate efficiently.
a 2:1 sampling rate.

B. PSNR Compensation IV. Simulation Results


The PSNR compensation step is introduced to improve A. Experimental Environment
the PSNR performance of the algorithm when the PSNR The overall structure of the proposed algorithm is given in
of the algorithm is much less that of the full search. The Fig. 3 and its performance was evaluated by implementing
PSNR compensation is conducted by executing three steps and comparing the FS algorithm, the MSEA, the CPST in [8]
(from the candidate subsampling step to the fine search step) with MSEA as the CRA, and various versions of the proposed
with a new search window that is centered at the final MV algorithm. Table II shows the implemented algorithms and
that was obtained from the previous fine search step. The their labels. When determining SADmax , the SAD values of the
PSNR compensation step is performed only if the final MV three MVs of the left, top, and top-right blocks of a current
that was obtained from the previous fine search step causes block are required. If any of these three SAD values is not
unacceptable PSNR performance degradation compared with available, we set the value of unavailable SAD at 2048. Also,
LEE et al.: 2:1 CANDIDATE POSITION SUBSAMPLING TECHNIQUE FOR FAST OPTIMAL MOTION ESTIMATION 1055

Fig. 3. Overall structure of the proposed algorithm.

TABLE III
Computations and PSNR Performance Comparison of Different Algorithms
Mobile (CIF) Foreman (CIF) Fun (CIF) Paris (CIF)
Total PSNR to FS Total PSNR to FS Total PSNR to FS Total PSNR to FS
FS 668516.3 0.000 668516.3 0.000 668516.3 0.000 668516.3 0.000
MSEA 33416.5 0.000 41878.2 0.000 38361.9 0.000 15265.40 0.000
Conv CPST 32919.3 −0.090 33540.9 −0.336 29052.0 −0.013 12237.9 −0.008
Prop CPST1 28986.4 −0.020 29248.2 −0.030 25521.9 −0.018 11488.9 −0.002
Prop CPST2 30999.7 −0.019 29345.5 −0.019 27157.9 +0.001 11707.2 +0.005
Proposed 20270.8 −0.027 24708.8 −0.018 21028.3 +0.003 10555.5 +0.001
Mobile (CCIR601) Football (CCIR601) Flower (CCIR601) Tennis (CCIR601)
Total PSNR to FS Total PSNR to FS Total PSNR to FS Total PSNR to FS
FS 2636526.3 0.000 2636526.3 0.000 2636526.3 0.000 2636526.3 0.000
MSEA 113322.0 0.000 216924.0 0.000 131796.0 0.000 823889.0 0.000
Conv CPST 137153.0 −0.074 145081.0 −0.010 106872.0 −0.023 496364.0 −0.194
Prop CPST1 73229.1 −0.014 126763.0 −0.016 78660.1 −0.014 459910.0 −0.054
Prop CPST2 77345.7 −0.010 140055.0 +0.073 82295.7 +0.002 482898.0 +0.004
Proposed 60985.1 −0.015 113452.0 +0.075 61163.8 +0.008 434055.0 +0.032

if any MVs in the three MVs of the left, top, and top-right minimum SAD value (Condition 3). The proposed condition
blocks of a current block for predicting the optimal MV is does not change the basic procedure in Algorithm 1, but is able
not available, it is replaced with (0, 0). All versions of the to find winners effectively. Adopting a new winner condition
proposed algorithm adopt MSEA as the CRA. yields a 16.05% reduction in computation and an improvement
The test platform for experiments was based on a simple in PSNR of +0.073 dB over Conv CPST. It is noticeable
code only to perform motion estimation. Search order for that the worst degradation of PSNR performance is reduced
candidate subsampling is in Fig. 1(a). Simulations were per- from −0.336 dB (Foreman in CIF) to −0.054 dB (Tennis in
formed in eight sequences with the two different formats: CIF CCIR601) by using a new winner condition.
(352 × 288, 50 frames, 25 frames/s) and CCIR601 (720 × 480, 2) PSNR Compensation (Prop CPST2): The PSNR Com-
30 frames, 30 frames/s) [11]. Only luminance components of pensation step is performed if SAD(MVFS ) is larger than
each sequence were used. The block size was set to 16 × 16. SADmax , in Fig. 3. Our simulation shows that the probability
The search range to ±15 for CIF and ±30 for CCIR601. of conducting the PSNR compensation step is 0.1358. It is
We set one operation of addition, subtraction, absolute value, natural to think that the PSNR compensation increases the
comparison, and shift each to equal one operation. computational load of Prop CPST1 by 13.58%. However, it
requires only 5.76% additional computations, giving a PSNR
B. Experimental Results improvement of +0.026 dB, because some area of the new
1) New Condition for Winner (Prop CPST1): Even search window overlaps with the original search window and
though Conv CPST subsamples candidate positions at a 2:1 the temporal minimum SAD value is already very small.
rate, an ineffective condition for choosing the winner results in It is remarkable that the PSNR compensation step reduced
slower performance. Conv CPST requires more computations the worst PSNR degradation from −0.054 dB (Tennis in
even than the MSEA for the Mobile sequence in CCIR601. CCIR601) to −0.019 dB (Mobile and Foreman in CIF).
This is because the fact that lower bounds are compared, not 3) Fast Motion Estimation (Proposed): The proposed
with the first minimum SAD, but with the second minimum algorithm applies the fast ME step to Prop CPST2 to reduce
SAD, while the MSEA always compares lower bounds with the number of computations by having the lower temporal
the first minimum SAD. minimum SAD value for the same candidate. On average the
Prop CPST1 only changes the condition for winner in proposed algorithm requires 15.38% fewer computations than
the Conv CPST (Condition 2 in Table I) into the proposed Prop CPST2. However, there is a small difference between
condition of whether or not a MV updates the temporal the PSNR performances of Prop CPST2 and the proposed al-
1056 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 20, NO. 7, JULY 2010

TABLE IV the Football sequence, speed gains yielded by the proposed


Computations and PSNR Performance Comparison of the ESEA algorithm over the MSEA and ESEA are, respectively, 47.70%
and the Proposed Algorithm and 46.69%. For Tennis, they are 47.32% and 46.68%.
Football (CCIR601) Tennis (CCIR601)
Total PSNR to FS Total PSNR to FS
V. Conclusion
ESEA 113131.0 0.000 376392.0 0.000
Proposed 60314.5 +0.075 200708.0 +0.032 A new 2:1 CPST that matches the MSEA has been
proposed. We proposed a new condition for choosing MVs,
which is used in the fine search step. This condition does
gorithm (+0.003 dB), because the different temporal minimum not need additional computation. Even so, it reduces the
SAD value causes different MVs to be selected as winners. probability that the CPST misses the optimal MV that is not
When comparing the MSEA with the proposed algorithm, in the subsampling pattern to 0.081. In order to deal with the
computational loads are reduced by 47.26% with +0.007 dB case in which the CPST misses the optimal MV, a PSNR com-
PSNR performance improvement on average. The minimum pensation step repeats three steps from the candidate position
speed gain is 30.85% for the Paris sequence in CIF, and the subsampling step to the fine search step with a new search
maximum speed gain is 53.59% for the Flower sequence in window. Moreover, the newly adopted fast ME step reduced
CCIR601. It is notable that the worst degradation in PSNR the number of computations required by the CPST by using
performance (Mobile in CIF) is only −0.027 dB worse than the the smaller temporal minimum SAD for the same candidate.
FS. This shows the proposed algorithm has PSNR performance Simulation results have shown that the proposed algorithm
that is comparable to that of the FS. produces significantly higher speedup gain over the MSEA
In five sequences (Fun and Paris in CIF; and Football, (47.26%) with +0.007 dB PSNR improvement for the tested
Flower, and Tennis in CCIR601), the proposed algorithm video sequences. This amount of performance improvement
exhibits higher PSNR performance than the optimal algorithms with regards to speedup gain and PSNR performance can
(FS and MSEA), which implies that some blocks in those se- be achieved regardless of the CRA if the CRA follows
quences are more similar to blocks outside the search window Algorithm 1.
than to blocks inside the search window. We examined the
simulation results for those sequences and found that the mean References
squared error (MSE) for the proposed algorithm was smaller
[1] T. Koga, K. Ilinuma, A. Hirano, Y. Iijima, and T. Ishiguro, “Motion-
than that of the optimal algorithms in two cases. In these two compensated interframe coding for video conferencing,” in Proc. Nat.
cases, better PSNR performance was achieved by shifting the Telecommun. Conf., 1981, pp. G5.3.1–G5.3.5.
position of a search window either after the fast ME step or [2] S. Zhu and K. K. Ma, “A new diamond search algorithm for fast block-
matching motion estimation,” IEEE Trans. Image Process., vol. 9, no.
before the PSNR compensation step in the proposed algorithm. 2, pp. 287–290, Feb. 2000.
The first case is related to large motions which are not [3] C. H. Cheung and L. M. Po, “Adjustable partial distortion search
covered by a search range, such as the motions in blocks algorithm for fast block motion estimation,” IEEE Trans. Circuits Syst.
Video Technol., vol. 13, no. 1, pp. 100–110, Jan. 2003.
containing information of balls in Paris and Tennis and players [4] X. Q. Gao, C. J. Duanmu, and C. R. Zou, “A multilevel successive
in Football. The second case is associated with bar-shaped elimination algorithm for block matching motion estimation,” IEEE
objects. If a block contains a part of a vertical or horizontal Trans. Image Process., vol. 9, no. 3, pp. 501–504, Mar. 2000.
[5] F. Moschetti and E. Debes, “A fast block matching for SIMD processors
bar-shaped object, occasionally a block outside the search using subsampling,” in Proc. IEEE ISCAS, vol. 4. May 2000, pp. 321–
window has a smaller MSE than any block inside the search 324.
window. A vertical or horizontal bar-shaped object tends to [6] C. N. Wang, S. W. Yang, C. M. Liu, and T. Chiang, “A hierarchical
N-queen decimation lattice and hardware architecture for motion esti-
appear in a series of blocks along the direction of bar-shaped mation,” IEEE Trans. Circuits Syst. Video Technol., vol. 14, no. 4, pp.
objects. Thus, a block outside the search window can have 429–440, Apr. 2004.
a smaller MSE by reducing errors corresponding to a bar- [7] R. Korah, M. Sankaralingam, and J. R. P. Perinbam, “Motion estimation
with candidate block and pixel subsampling algorithm,” in Proc. IEEE
shaped object. The bar-shaped objects in Fun are the bars of a Int. Workshop IST, May 2005, pp. 130–133.
merry-go-round and the bar-shaped objects in Flower are the [8] W. Jiang and M. Zhou, “A fast BMA based on combining search
branches and logs of a tree. candidate subsampling and APDS,” in Proc. IEEE ICME, vol. 2. Jun.
2004, pp. 1115–1118.
The proposed algorithm does not modify the MSEA itself. [9] A. M. Tourapis, O. C. Au, and M. L. Liou, “Highly efficient predictive
This implies that the proposed algorithm can be combined with zonal algorithms for fast block-matching motion estimation,” IEEE
any of the fast full-search algorithms following Algorithm 1, Trans. Circuits Syst. Video Technol., vol. 12, no. 10, pp. 934–947, Oct.
2002.
so we can expect that the proposed algorithm will reduce the [10] H. S. Lee, J. H. Jung, and D. J. Park, “An effective successive elim-
computational load of the fast full-search algorithm by 47% ination algorithm for fast optimal block-matching motion estimation,”
with the almost same PSNR performance. The effective suc- in Proc. IEEE ICIP, Oct. 2008, pp. 1984–1987.
[11] Test Image Sequences. [Online]. Available: http://www.info-
cessive elimination algorithm (ESEA) is an improved version ab.uclm.es/asignaturas/42609/Material.htm
of the MSEA [10]. For two sequences in CCIR601, Football
and Tennis, the proposed algorithm that has the ESEA as
the CRA reduces computations of the ESEA by 46.69% and
46.68% with improvements in PSNR of +0.075 dB and +0.032
dB, respectively. Note that the proposed algorithm yields the
almost same speed gain over the ESEA and the MSEA. For

You might also like