A STUDY ON MEDIA SCALING OF MULTI-VIEW MOTION PICTURES

MPEG2是一种标准的电影编码方法，由于标题等所耗费的信息，在MPEG2中很难减少编码信息量。在立体电影中，我们知道，如果一只眼睛看到高分辨率的图像，人们不会注意到另一只眼睛看到的图像的退化。利用上述特性，可以在不被观察者注意到的情况下对多视点运动图像进行压缩和传输。本课题提出了一种改进的同播编码方法，即利用MPEG2对多视点运动图像中的一个位置进行编码。其余图像对参考图像进行视差补偿，剩余图像以低空间和时间分辨率进行编码。与MPEG2相比，编码这些图像所需的信息量约为十分之一。该方法可以在与传统单通道节目相同的带宽下播放立体图像。

It is difficult to reduce the amount of encoded information in MPEG2, which is a standard encoding method for motion pictures, because of the information spent by headers and so on. In stereo motion pictures, it is known that if a high-resolution picture is presented to one of the eyes, human does not notice the degradation of the picture presented to the other eye. Using above-mentioned characteristics, it is possible to compress and transmit the multi-view motion pictures without noticed by the observer.In this research project, modified simulcast encoding method in which one of the positions in multi-view motion pictures is encoded by MPEG2 is proposed. Parallax compensation against reference pictures are applied to the other pictures, and the residual pictures are encoded in low space and time resolution. The amount of the information required to encode these pictures is about one tenth compared with MPEG2.The method can be applied to broadcast stereo pictures in the same bandwidth as conventional single channel program.