基于低秩张量恢复的视频块效应处理

针对块编码的视频解码后存在块效应的问题,提出了一种基于块和低秩张量恢复的块效应处理方法。首先在视频序列里寻找相似块构造三阶张量,根据背景张量的低秩性和块效应的稀疏性,利用扩展于张量上的增广拉格朗日乘子法求解一个低秩张量恢复问题。从张量模型的角度来进行视频块效应处理,更好地保护了高维数据的结构特性。实验结果显示,相对于传统去块效应方法,通过该方法得到的视频图像有更高的峰值信噪比(PSNR)和更好的视觉效果。

Block-coded Video Deblocking Based on Low-rank Tensor Recovery

Block-coded videos suffer from the blocking artifacts after being decoded.In order to solve this problem,a block-based deblocking method using low-rank tensor recovery was proposed.First,three order tensor is constructed through clustering similar blocks in video sequence.Then,according to low-rank property of background tensor and sparsity of blocking artifacts,the proposed approach utilizes the augmented Lagrange multiplier method which extends to tensor to solve the low-rank tensor recovery problem.The proposed approach utilizes tensor model to preserve the structural properties of high dimensional data.Experimental results show that it can obtain higher PSNR value and better visual effect comparing with traditional deblocking methods.