extract rectangular area from noisy image

Question

extract rectangular area from noisy image

I want to get the clean figure (right) from the noisy figure (left):

To do this, I have a dataset with the noisy figure and its corresponding clean figures. When I put the algorithm into a noisy figure, I want it to estimate the clean figure. Which machine learning algorithm should I use to achieve this?

I know I will use Python, but I am unsure how to input this dataset and to which algorithm? I would appreciate any help provided.

Answer 1

你可以把它当作一个分割任务，并为它使用标准的 U-Net 架构——PyTorch 和 TensorFlow 中有很多实现