Bracketing Image Restoration and Enhancement with High-Low Frequency Decomposition-高频分解

paper
通过密集连接的小卷积核实现细节特征（高频特征提取）提取

import torch.nn as nn
class Dense(nn.Module):
    def __init__(self, in_channels):
        super(Dense, self).__init__()

        # self.norm = nn.LayerNorm([in_channels, 128, 128])  # Assuming input size is [224, 224]
        self.conv1 = nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1,stride=1)
        self.conv2 = nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1,stride=1)
        self.conv3 = nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1,stride=1)
        self.conv4 = nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1,stride=1)
        self.conv5 = nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1,stride=1)
        self.conv6 = nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1, stride=1)

        self.gelu = nn.GELU()

    def forward(self, x):

        x1 = self.conv1(x)
        x1 = self.gelu(x1+x)

        x2 = self.conv2(x1)
        x2 = self.gelu(x2+x1+x)

        x3 = self.conv3(x2)
        x3 = self.gelu(x3+x2+x1+x)

        x4 = self.conv4(x3)
        x4 = self.gelu(x4+x3+x2+x1+x)

        x5 = self.conv5(x4)
        x5 = self.gelu(x5+x4+x3+x2+x1+x)

        x6= self.conv6(x5)
        x6 = self.gelu(x6+x5+x4+x3+x2+x1+x)

        return x6