An Equilibrium Technology Between Foreground And Background In Video Sequence

With the rapid development of IT, the technology of video coding becomes more and more important. Video phone and video meeting certainly will become more and more popular, but each of them has a great deal of data to be transferred. At the present time, the bandwidth of network that can be used is not enough to satisfy our need. At the limit of the bandwidth of network, we have to compress the data of the video sequence to assure the output bits of the system are in the anticipant range. In most mainstream coding/decoding systems, such as H.263 and H.26L, they treat equally each part of the picture to be coded. Thus it ignores the subjectivity of the men's vision. We often expect that the video quality of the interesting region could be better. About the others that we have no interest in, we don't mind their quality very much.Firstly, this topic puts forward a new resolution to original Interlace algorithm of H.26L which is widely adopted in the mainstream coding/decoding systems. Then, it presents a series of technologies to implement foreground_background segmentation and bits_assigned on foreground and background. The limitation of the original Interlace algorithm in H.26L is that it can't match with reality scene very well. In order to be applicable all most reality scene, we expand the matching pattern of the algorithm. The significance of the amelioration is more availably information can be transferred in the same bits. In order to make the output video sequence good quality, we implement foreground_background segmentation at the platform with the ameliorated algorithm. We primarily adopt the technology of skin segmentation and motion segmentation, then we fill the picture that have been segmented. At the base of foreground_background segmentation, we put the bits that originally belong to the background region to the foreground region. So the video quality in the foreground region is increased greatly. As well as the video quality in the background region is reduced. But analyzing the result in the whole, we find that the encoding system not only improve the compress ratio of the video sequence but also can assign the code bits appropriately. So we can get better subjective video quality at the limit of low bits ratio.