将图层亮度与图层色度相结合(Combine layer chrominance with image luminance)

将图层亮度与图层色度相结合(Combine layer chrominance with image luminance)

我已经阅读了这本彩色纸 ，它说：

彩色化网络的输出层由一个具有Sigmoid传递函数的卷积层组成，该函数输出输入灰度图像的色度。

为了得到彩色图像，他们说：

计算出的色度与输入强度图像组合以产生最终的彩色图像。

所以我已经实现了它并获得了深度为2的输出图层，但是如何获取彩色图像呢？ 如何将灰度图像亮度值与深度2的输出图层（a * b颜色）组合以获得最终图像？

我使用tensorflow和python。

I have read this Colorization paper and it said:

The output layer of the colorization network consists of a convolutional layer with a Sigmoid transfer function that outputs the chrominance of the input grayscale image.

and in order to get the colored image they said:

the computed chrominance is combined with the input intensity image to produce the resulting color image.

So I have implemented it and get the output layer with depth two, but how can I get the color image? How can I combine the greyscale image luminance values with the output layer of depth 2 (a*b colors) to get the final image?

I use tensorflow and python.

最满意答案

好吧，我试图通过使用Skimage库来实现它，以制作图像色度值的张量并用相同的方法将其与亮度进行比较。

Ok, I tried to implement it by using Skimage library to make a tensor of image chrominance values and compine it with the luminance by the same method.