This paper describes a fast digital video (DV) to MPEG-2 conversion algorithm. At first, DCT domain 2/spl times/4/spl times/8 DCT to 8/spl times/8 DV conversion was employed for the DV to MPEG-2 I-picture conversion. DV to MPEG-2 IBP-picture conversion requires motion estimation, so fast hierarchical motion estimation was also employed. Since DCT information is available from DV data and 2/spl times/2 IDCT can be achieved by simple addition and subtraction operations, a low-resolution image is constructed and used for a coarse motion search. Then a fine search is performed using extended 2SS in the original image domain. The experimental results show that the proposed method is more than 40 times faster than a baseband domain conversion with the full search while about the same PSNR is obtained. Both the PSNR and the conversion speed are also greatly improved when compared with the conventional method.
[1]
Yoonsik Choe,et al.
Conversion between DV and MPEG-2 intra coding
,
2001,
ICCE. International Conference on Consumer Electronics (IEEE Cat. No.01CH37182).
[2]
Wen-Hsiung Chen,et al.
A Fast Computational Algorithm for the Discrete Cosine Transform
,
1977,
IEEE Trans. Commun..
[3]
Akio Yoneyama,et al.
Fast dissolve operations for MPEG video contents
,
2000,
Proceedings 2000 International Conference on Image Processing (Cat. No.00CH37101).
[4]
C.C. Lee,et al.
New motion estimation algorithm for DV to MPEG1 compression domain conversion
,
1999,
1999 Digest of Technical Papers. International Conference on Consumer Electronics (Cat. No.99CH36277).