A new motion activity measure in temporal video transcoding

Transcoding is an important technique for adapting the bit rate of previously compressed video sequences to different network bandwidths. Frame skipping is the type of transcoding often used to reduce the frame rate and to allocate more bits to more representative frames in order to guarantee an acceptable quality of the whole video sequence. The motion activity is the most important factor in the choice of frames to be skipped. Traditionally, usual schemes consider the sum of motion vectors of the current frame as measure of motion activity. Frames with a large motion activity are transcoded to preserve a smooth motion. In this paper, we propose a new way to compute the motion activity measure able to consider different kinds of motion in each frame. Simulation results show that this new measure achieves a better performance than the standard one.