摘要
提出一种实现视频会议中前后景分割和比特数分配的技术 .在帧间预测的基础上 ,对序列图像中的运动目标或感兴趣区域 (视频会议中为人脸区域 )进行检测和跟踪 ,从而实现前后景的分割 .该技术可以进一步提高图像数据的压缩比 ,而且还可以更合理地分配码流 ,能在较低比特率的限制下获得更好的主观视频质量 .
This paper proposed a technique to realize the foreground and background segmentation and bits assignment in video meeting. Based on inter frame prediction, the foreground can be segmented from the original image by detecting and tracking moving target or interesting area in image sequence. In a lower rate of bits case, the foreground is assigned more bits to encode to produce a good quality subjectively. It can raise the compression ratio and assign the code flow.
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2004年第5期11-13,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
航天技术创新基金项目
关键词
图像分割
视频会议
均衡技术
运动检测
肤色检测
image segmentation
video meeting
equilibrium technology
motion detection
skin detection