MPEG video coding is a kind of compression technique which exploits intra frame and inter frame infomation of video signal, for example, predictive coding, transform coding such as the discrete cosine transform(DCT), variable length coding, run length coding, motion estimation and compensation, etc. Since the MPEG video coding mainly use the predictive coding technique, one sends or stores the encoded pictures according to the transmission order which is different from the real displaying order. Because of the mismatch between the transmission order and the display order of the MPEG-encoded video sequences, we can edit the video signal under several limited conditions.
In this thesis work, we propose a moving picture editing algorithm which can support three kinds of picture types such as intra(I), predictive(P), and bidirectionally(B) coded pictures. The proposed algorithm realizes sequence editing independent of GOP (Group-Of-Picture) structure and irregularity of array of pictures in GOP. It does not require the synchronization of sequences to be edited either. We evaluate the performance of the proposed editing algorithm using PSNR(Peak-to-peak Signal to Noise Ratio) by comparing the edited sequence with the original raw sequence and the conventionally edited one. The results show that the performance of the proposed algorithm is superior to the conventional editing technique.
