欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对现有的大多视频修复方法为了满足视觉一致性而需要额外处理过程的问题,提出了一种基于时空几何流的Bandelet稀疏正则化方法。首先,利用Bandelet变换提取视频时空几何流特征以重建丢失的数据;然后,通过基于Bandelet块融合增强的优先级样本方法生成初步修复结果;最后,运用稀疏正则化完成剩余的修复任务。实验结果表明,相比其他几种较好的视频修复算法,该方法取得了更好的修复质量,并在均方差（MSE）衡量和时空最明显失真模型（STMAD）的时间一致性方面取得了更好的性能。
英文摘要：
For the issue that most existing video inpainting methods require additional processing due to visual consistency, this paper proposed a Bandelet sparsity regularization method based on spatio-temporal geometric flows. Firstly, it used Bandelet transformation to extract video spatio-temporal geometric flows features so as to reconstruct lost data. Then, it used the priority sample method based on Bandelet block enhancing to generate primary inpainting results. Finally, it used sparsity regularization to finish the rest task. Experimental results show that proposed method has better video inpainting quality than several other advanced algorithms. Also, it has better performance on the time measured in conformity assessment MSE and the space-time measurable targets to assess the most obvious distortion model （STMAD）.
同期刊论文项目
同项目期刊论文
Energy-efficient task scheduling model based on MapReduce for cloud computing using genetic algorith
A progressive multiple criteria sorting approach based on utility function considering imprecise inf
The Current State and Future Directions of Research and Practice in Organizational Learning and Lear
期刊信息
位置：