基于位差预测的视图合成

View Synthesis Based on Disparity Prediction

为了满足多视点视频中新视点合成的需要,提出了一种基于位差估计和补偿插值的视点合成方法。实验结果表明,当两个摄像机间的基线比较小的时候,合成的视图质量和传统方法相当,但合成速度比较快,可应用于实时的交互式系统中。

In order to synthesis the interview in the multi-view video,a new view synthesis method is proposed which is based on disparity prediction and interpolation.Experimental results show that the proposed method can provide good image quality compared with traditional methods when the baseline is small.The proposed method can get the intermediate view quickly thus it can be applied in interactive system.