| 基于低秩矩阵补全的视频去噪研究 |
 点此下载全文 |
| 引用本文：包文瑞.基于低秩矩阵补全的视频去噪研究[J].计算技术与自动化,2019,(2):90-95 |
| 摘要点击次数: 866 |
| 全文下载次数: 0 |
|
|
| 中文摘要:针对视频数据中严重的混合噪声问题，提出了一种基于块的视频去噪算法。通过对空间域和时间域的相似块进行分组，将混合噪声的去噪问题转化为一个低秩矩阵补全问题，从而得到一个对噪声统计特征没有强假设的去噪方法。由此产生的核范数最小化问题通过拉格朗日函数和不动点迭代算法得到有效的解决。实验验证了所提出的视频去噪方法在去除混合噪声方面的鲁棒性和有效性。 |
| 中文关键词:矩阵补全 低秩矩阵 视频去噪 核范数 不动点迭代 |
| |
| Video Denoising Based on Low Rank Matrix Complementation |
|
|
| Abstract:Aiming at the serious mixed noise problem in video data，this paper proposes a block-based video denoising algorithm. By grouping similar blocks in the spatial domain and the time domain，the denoising problem of mixed noise is transformed into a low rank matrix completion problem，and a denoising method with no strong assumptions on noise statistical features is obtained. The resulting kernel norm minimization problem is effectively solved by the Lagrangian function and the fixed point iterative algorithm. The experiment verifies the robustness and effectiveness of the proposed video denoising method in removing mixed noise. |
| keywords:matrix complement low rank matrix video denoising kernel norm fixed point iteration |
| 查看全文 查看/发表评论 下载pdf阅读器 |
位访问者,今日访问
