BLOCK MOTION ESTIMATION USING DIRECTIONAL ADAPTIVE SEARCH WINDOW

Motion estimation (ME) is the exploitation of similarities between adjacent frames in
a video sequence by eliminating temporal redundancy, and is an essential part of the
H.264 and other video compression standards. However, it introduces an increase of
computation complexity resulting in longer execution time. Thus, adaptive motion
estimation for H.264 is proposed in order to reduce the execution time while giving
better PSNR performance. The algorithm determines the amount of motion in each
block and classifies them as low, medium and high motion. From the magnitude and
direction of the x andy motion vector components, the search window (search range)
is dynamically adjusted. For high motion, the search range is set to be the maximum
value and vice versa. The results show that execution time could be reduced to almost
half (50%) of the conventional method since the number of search points and
computations decrease inthe range of40% to 60%. Furthermore, the method gives a
better image quality for video sequence with uniform motion and negligible PSNR
loss in others. By introducing early termination inthe adaptive motion estimation, the
number of computation could be reduced even further since the search process is
terminated immediately certain criteria are satisfied. By using Option 2 for early
termination, the search point computation and PSNR is reduced with average 1.3%
and 1.027% from the adaptive motion estimation without the early termination
process.