Biography

Multi-Stage Interval-Based Motion Estimation (MIME) Algorithm

Description
Multi-Stage Interval-Based Motion Estimation (MIME) Algorithm
Categories
Published
of 23
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Multi-Stage Interval-Based Motion Estimation (MIME)Algorithm Hanan Mahmoud, Sumeer Goel, Mohsen Shaaban, and Magdy Bayoumi  Abstract  This paper presents a new full-search block-matching algorithm: Multi-stage Interval-based Motion Estimation algorithm (MIME). The proposed algorithm reduces the computational load by successively eliminating non-candidate blocks from the search window. Thiscomputational reduction leads to enhanced performance in terms of low power consumption and  fast motion vector estimation. A low power VLSI implementation of the algorithm is also presented in this paper. Simulation results on benchmark video sequences are presented. 1. INTRODUCTION Video compression aims at compressing the amount of data necessary to transmit a videosequence across a bandwidth-limited channel. Motion estimation is considered as the mostcomputational expensive operation in any video codec. Motion estimation aims at reducing thetemporal redundancy between successive frames in a video sequence [1-6]. Motion analysistechniques are used to generate motion vectors that are transmitted instead of the actual framedata. One such popular technique for motion estimation is the block-matching algorithm (BMA)[7]. In this technique, the current image frame is first partitioned into fixed-sized rectangular  blocks, and the motion vectors for each block is estimated by finding the best matching block of  pixels in the previous frame according to a matching criterion. Full-search block-matching  (FSBM) algorithm [8] employs this technique. FSBMA provides optimum performance bysearching all the blocks in the search window. Since FSBMA searches all the blocks in thesearch window, it is computationally expensive and limits its practical applications. The power consumption and computational cost of these search algorithms can be reduced at different levelsof abstraction. Several cost effective techniques at algorithmic level have been proposed in theliterature [9-12]. Besides these, enhancements at circuit level can also be incorporated [13-14].These modifications address the problem of power consumption but compromise on thecomplexity of the approach.In this paper, we present an enhancement to the present FSBM algorithm that reduces thealgorithmic complexity as well as power consumption. Our approach is based on successiveelimination [15] of candidate blocks from the search window using an approximate interval bounding the distortion value i.e. the SAD . The two boundaries of the interval are two novelfunctions that are approximations of the actual SAD function. The calculations of theseapproximate functions are inexpensive in comparison with the actual SAD calculation and thusreduce the computational load drastically. In the next section, we discuss the FSBM algorithm.Section 3 discusses the FSBM algorithm based on conservative approximation. We present thenew Multi-Stage Interval-Based Motion Estimation (MIME) Algorithm in section 4 and the proposed low-power VLSI architecture is presented in section 5. Simulation results are discussedin Section 6. II. FULL-SEARCH BLOCK-MATCHING ALGORITHM Full-search block matching algorithm (FSBMA) finds the best match for each reference block of size  N  x N  in the current frame within a search area S  in the previous frame. The  criterion for best match is the candidate block with the minimum amount of distortion whencompared with the reference block. The most common measure used for calculating distortion isthe sum of absolute differences ( SAD) of intensity values between the two blocks beingcompared. The SAD for the candidate block of size  N  x N  at position (u,v) can be defined as: ∑∑ = = −++=  N i N   j  jir v jui svuSAD 1 1 ),(),(),( (1)where ),( jir  and ),( v  jui s ++ are intensity values at position ),( ji of the reference block and),( v jui ++ of the candidate block in search area S  . The search area is formed by extendingthe reference block by a search range w on each side (refer to Fig. 1) forming a search area of  (2w+N) 2    pels. As a result, there are (2w +1) candidate blocks in both horizontal and verticaldirections i.e. a total of  (2w+1) 2   candidate blocks have to be searched corresponding to eachreference block. The distortion value is computed for each candidate block and the minimumvalue min SAD is found from the pool of  (2w+1) 2 candidates. The block matching processgenerates a motion vector  min ),( vu and the corresponding distortion value min SAD . FSBMA iswidely used because of its simplicity and regularity, but it needs massive computations and anexpensive hardware. III. BMA BASED ON CONSERVATIVE APPROXIMATION This algorithm [16] is based on successive elimination principle [15] and makes aconservative approximation of the distortion function ),( vuSAD for the estimation of motionvectors. The calculation of the new estimate ),( vu D is relatively less expensive in terms of  power consumption as compared to the computation of the conventional ),( vuSAD . Theconservative estimate of the ),( vu D is given as:  ∑∑ ∑ −= = = −++= 11 1 1 ),(),(),(  N i N   j N   j   jir v  jui svu D (2)The new function ),( vu D proves to be a lower bound of the function ),( vuSAD .Initially, ),( baSAD for any random location (a,b) in the search window is computed and is setas minimum distortion so far ( min  D ). Thereafter the conservative estimate ),( vu D iscomputed for all remaining candidate blocks. If the conservative estimate for a candidate block islarger than the minimum distortion so far  min  D , then that candidate block is eliminated or notconsidered as a candidate for matching i.e. there is no need to compute the exact distortion. If theconservative approximate is less than the min  D then min  D is replaced by this conservativeapproximate and the candidate block is put in a set of candidate blocks whose actual SAD will becalculated. This is repeated for all candidate blocks in the search area S  . The saving in power iscontributed to the eliminated candidate blocks as long as the power consumed to calculate theconservative estimate is less than that consumed for calculating the exact distortion.Careful analysis of equation (2) shows that the conservative estimate ),( vu D is notdirectly proportional to the exact distortion ),( vuSAD thereby limiting the capability of thealgorithm. This can be proved by an example shown in Figure 2. Four blocks of 4 x 4 pixels areshown here. For simplicity, only two possible pixel intensity values are taken.In this example, exact distortions and the conservative approximates are calculated for various blocks. It is found that ),(),( d aSADbaSAD > but ),(),( d a Dba D < . According tothe conservative approximation algorithm, this candidate block will not be eliminated althoughits SAD value suggests that it should have been eliminated. The same can be observed for block a, block c and block d where ),(),( d aSADcaSAD > but ),(),( d a Dca D < . This proves thatthe conservative approximate is not directly proportional to the exact distortion. As a result of   this discrepancy there is lesser number of candidate blocks eliminated from the search area i.e.the exact distortion will have to be calculated for more number of candidate blocks. Figure 3shows the average percentage distribution of blocks where ),( vu D is proportional to ),( vuSAD for different benchmark video sequences.Another observation we made is that the number of blocks eliminated by the algorithmdepends heavily on the choice of the starting point because the exact distortion is calculated for the starting point and set to min  D and for the remaining points the conservative approximate iscalculated and compared to this min  D . Figure 4 shows the average number of blocks eliminatedusing conservative approximation with different starting points for various benchmark videosequences. IV. THE PROPOSED ALGORITHM We propose the multi-stage interval-based motion estimation (MIME) algorithm. The proposed algorithm is a block based motion estimation algorithm that utilizes successiveelimination technique. We define two approximate functions, ),( )(1 vuSAD m and ),( )(2 vuSAD m as the upper and lower boundaries, respectively, of the interval that includes ),( vuSAD . The character  ‘m’  is equal to 2 b1 where ‘b1’  is the number of the bits used in the pixel intensity starting from the  MSB going to the  LSB . For example, we can use only two MSBsof the pixel intensity value for both, current and reference frame, to calculate ),( )4(1 vuSAD and ),( )4(2 vuSAD .
Search
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x