Learning a Deep Multi-Scale Feature Ensemble andan Edge-Attention Guidance for Image F（红外图像和可见光图像1）

一、Learning a Deep Multi-Scale Feature Ensemble and
an Edge-Attention Guidance for Image Fusion（基于深度多尺度特征集成的图像融合边缘注意引导）

代码分成了三个部分

（1）：Extractor 

def __init__(self):
    super(Extractor, self).__init__()

    # group S
    self.conv_1 = ConvBlock(1, 16, p=1)

    # group A
    self.conv_a1 = ConvBlock(16, 32, p=1)
    self.conv_a2 = ConvBlock(32, 48, p=1)
    self.conv_a3 = ConvBlock(48, 64, p=1)

    # group B
    self.conv_b1 = ConvBlock(16, 32, p=2, d=2)
    self.conv_b2 = ConvBlock(32, 48, p=1)
    self.conv_b3 = ConvBlock(48, 64, p=1)

    # group C
    self.conv_c1 = ConvBlock(16, 32, p=3, d=3)
    self.conv_c2 = ConvBlock(32, 48, p=1)
    self.conv_c3 = ConvBlock(48, 64, p=1)

(2):Constructor

self.conv_1 = ConvBlock(64, 16, p=1)
self.conv_2 = ConvBlock(64, 32, p=1)
self.conv_3 = ConvBlock(64, 16, p=1)
self.conv_4 = ConvBlock(16, 1, p=1)

(3):边缘导向注意机制的过程

self.conv_1 = ConvConv(1, 32, 32)
self.conv_2 = ConvConv(32, 64, 128)
self.conv_3 = ConvConv(128, 64, 32)
self.conv_4 = nn.Conv2d(32, 1, (1, 1))

self.ed = EdgeDetect()