胶片电影中的快速斑块修复算法研究  

Research on Fast Patch Repair Algorithm in Film Movie

在线阅读下载全文

作  者:刘泉洋 刘云清[1] 史俊 颜飞[1] 张琼 LIU Quan-yang;LIU Yun-qing;SHI Jun;YAN Fei;ZHANG Qiong(School of Electronics and Information Engineering,Changchun University of Science and Technology,Changchun 130022)

机构地区:[1]长春理工大学电子信息工程学院,长春130022

出  处:《长春理工大学学报(自然科学版)》2021年第4期57-63,共7页Journal of Changchun University of Science and Technology(Natural Science Edition)

基  金:吉林省科技厅重点项目(20190303080SF);吉林省科技发展重点研发项目(20190303034SF)。

摘  要:斑块是胶片电影中最常见的损伤类型问题,传统的斑块修复算法修复质量较好,但修复速度不够快。针对传统斑块修复算法在修复速度上的不足,提出了一种改进性的快速斑块修复算法。相比于传统修复算法的改进主要体现在运动估计模块和修复模块。运动估计模块使用权重系数矩阵对二维匹配块矩阵做降维处理,得到两组一维特征矩阵,使用一维特征矩阵代替二维匹配块搜索运动矢量最优值。修复模块使用单参考帧修复算法代替双参考帧修复算法。实验结果表明,快速斑块修复算法与传统修复算法相比,主观评价基本一致,修复速度优于传统修复算法。Patches are the most common type of damage in film.Traditional patch repair algorithms have better repair quali-ty;however,the repair speed is not fast enough.Aiming at the shortcomings of traditional patch repair algorithms in repair speed,an improved fast patch repair algorithm is proposed.Compared with the traditional restoration algorithm,the im-provement is mainly reflected in the motion estimation module and the restoration module.In the motion estimation module,the weight coefficient matrix to perform dimensionality reduction processing on the two-dimensional matching block matrix is used to obtain two sets of one-dimensional feature matrices;and the one-dimensional feature matrix is used to replace the two-dimensional matching block to search for the optimal value of the motion vector.A single reference frame repair algorithm is used instead of a dual reference frame repair algorithm.The experimental results show that being compared with the traditional repair algorithm,the subjective evaluation of the fast patch repair algorithm is basically the same and the repair speed is better than the traditional repair algorithm.

关 键 词:胶片电影 运动估计 块匹配算法 运动矢量 参考帧 

分 类 号:TP391.41[自动化与计算机技术—计算机应用技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象