出版社：The Institute of Image Information and Television Engineers
摘要：Recent years have seen the development of methods for motion-compensated television standards conversion (MC-TSC) in which motion information in the picture is utilized for conversion between TV standards that have different numbers of scanning lines and fields. Motion information used for standards conversion should contain true movement, which is not always a requirement in motion-compensated predictive coding. In conventional motion estimation methods for MC-TSC, however, the trueness has been insufficient, which has led to degradation of the converted picture. The authors therefore have improved the motion estimation scheme, utilizing a DOG (Difference of Gaussian) preprocessing filter for motion estimation, the iterative gradient method with improved high-precision, and the AOSF (Absolute Order Statistics Filter) for post-processing of motion vectors. A new high-quality motion-compensated television standards converter has been developed adopting the improved motion estimation scheme. This converter now makes it possible to greatly suppress degradation of the converted picture which has been difficult up until now.
