among the four aforementioned components, the choices of propagation and alignment components could lead to a big swing in terms of performance and efficiency.
在上述四个组件中，传播和对齐组件的选择可能会导致性能和效率的大幅波动。
propagation, alignment, aggregation, and upsampling. 传播、对齐、聚合和上采样
the use of bidirectional propagation scheme to maximize information gathering, and an optical flow-based method to estimate the correspondence between two neighboring frames
for feature alignment. By simply streamlining these propagation and alignment components with the commonly adopted designs for aggregation (i.e. feature concatenation)
and upsampling (i.e. pixel-shuffle [27]), BasicVSR outperforms existing state of the arts [9, 12, 32] in both performance (up to 0.61 dB) and efficiency (up to 24× speedup).
我们的实验建议使用双向传播方案来 最大化信息收集，并使用基于光流的方法来估计两个相邻帧之间的对应关系以进行特征对齐。通过使用普遍采用的聚合（即特征连接）和上采样（即像素混洗 [27]）设计简单地简化这些传播和对齐组件，
IconVSR : propagation, aggregation 传播和聚合
- information-refill.
This mechanism leverages an additional module to extract features from sparsely selected frames (keyframes), and the features are then inserted into the main network for feature refinement.
这种机制利用一个额外的模块从稀疏选择的帧（关键帧）中提取特征，然后将特征插入到主网络中进行特征细化。
- a coupled propagation scheme
which facilitates information exchange between the forward and backward propagation branches.
第二个扩展是耦合传播方案，它促进了前向和反向传播分支之间的信息交换。
The two modules not only reduce error accumulation during propagation due to occlusions and image boundaries, but also allow the propagation to access complete information in a sequence for generating high-quality features.
这两个模块不仅减少了传播过程中由于遮挡和图像边界引起的误差积累，但也允许传播访问序列中的完整信息以生成高质量的特征。
1）传播， 双向传播 ，他说的是 正向传（信息从第一帧到最后一帧顺序传播） + 反向传，
2）对齐， BasicVSR 采用简单的基于流的对齐，但发生在特征级别。
对齐从图像级别移动到特征级别会产生显着的改进。
将对齐的特征传递给多个残差块进行细化。
3）聚合和上采样， 特征连接和 pixelshuffle 的流行选择就足够了。
由多个卷积和像素混洗 [27] 组成的上采样模块用于生成输出 HR 图像：
IconVSR
– Information-refill mechanism and coupled propagation (IconVSR)
—信息重新填充机制和耦合传播 (IconVSR)，
1）Information-Refill. 信息补充。
feature refinement. 特征细化