基于运动估计的深度信息生成技术研究

Generation Technique Research of Depth Information Based on Motion Estimation

国内外的多家电视媒体和拥有立体显示技术的企业都进行过开播数字立体频道的尝试,立体电视(3DTV)是未来的一个趋势。而立体视频片源制作费用高,周期长,是目前亟需解决的问题。描述了一种3DTV解决方案,提出了一个平面视频转换立体视频的方法。并主要针对方法中的深度信息生成问题,结合运动估计算法,解决了小范围内非关键帧深度信息自动生成问题,简化了转化过程中的人为工作量,提高了效率,降低了制作成本。证明了单目平面视频转换为基于深度信息的立体视频的可行性,得到较为满意的结果。

Many TV media and companies having a three-dimensional display technology have been trying to launch digital stereo channels,3DTV is a trend of future.However,the high production costs and cycle length of the source of 3D video,is a urgent need to resolve.Described a solution for 3DTV,proposed a method of converting plane video to 3D video.For the generation of depth information in the method,solved a problem of automatically generating the depth information of non-key frame combinated with motion estimation algorithm.Simplified the human workload in convertion process,improved efficiency and reduced production costs.Proved the feasibility of conversion from 2D to 3D video and received a more satisfactory results.