极低码率下的2D-3D人脸视频编解码  

2D-3D Facial Video Coding/Decoding at Ultra-low Bit-rate

作　　者：於俊[1] 汪增福[1] 

机构地区：[1]中国科学技术大学自动化系,安徽合肥230027

出　　处：《电子学报》2013年第1期185-192,共8页Acta Electronica Sinica

基　　金：国家自然科学基金(No.60875026;No.60805019);安徽省科技攻关计划语音专项(No.11010202192);中国博士后科学基金(No.2012M521248)

摘　　要：针对动态变化背景下的人脸视频编解码问题,该文提出了一种2D-3D混合编解码系统;具体包括:(1)基于多种观测信息,在线外观模型和粒子滤波的人脸三维运动跟踪;(2)结合参数模型与肌肉模型的3D人脸动画合成;(3)基于头发检测和3D头发模型的头发合成;(4)无缝地拼接前景的三维编解码结果和背景的二维编解码结果.在极低码率下,客观实验表明,该系统在编码效率和解码质量上有较好的综合优势.主观实验表明,该系统的解码结果在脸部具有较高的辨识度.For facial video coding/decoding with dynamic background,a 2D/3D mixed system was proposed,it includes：（1）multi-measurements and online appearance models were applied to track 3D facial motion from video in framework of particle filter;（2）3D facial animation was produced combining parameterized model and muscular model;（3）3D hair was synthesized based on hair detection and 3D hair model;（4）3D coding/decoding result of foreground and 2D coding/decoding result of background were stitched seamlessly.At ultra-low bit-rate,the object experiment confirmed the comprehensive advantage between coding efficiency and decoding quality of this system,the between subjects experiment indicated the suitability of subjective face identification by it.

关 键 词：视频编解码 人脸运动跟踪 人脸动画 头发分析与建模 

分 类 号：TB391.9[一般工业技术—材料科学与工程]