| Summary: | 碩士 === 國立臺灣科技大學 === 資訊管理系 === 91 === Half--pel motion estimation is an important
ingredient in low bit rate video compression standards.
According to our new observation on the statistical data under three
video sequences, this paper presents a simple and efficient
prediction--based algorithm
for half--pel motion estimation. Experimental results reveal that
our proposed algorithm has better image quality in terms of
peek--to--signal noise ratio (PSNR)
while only having little execution--time degradation when compared to
the previous two algorithms, namely the linear interpolation method
by Jeong and Ahn, and the quadratic fitting method by Li and
Gonzales. Our proposed method has better PSNR and execution--time
performance when compared to the approximate criterion method by
Senda. Further, plugging our proposed prediction scheme into the
currently published method by Panusopone and Baylon, it leads to have
better execution--time performance and have similar PSNR.
|
|---|
