About the XDoG algorithm

[cating341]

Hi,

I am interested in this project. I tried to implement the same edge detection filter, XDoG, to extract the line from the colored illustrations. However, I can't get results as good as in your paper shown. Maybe there is something wrong in my code or my background knowledge isn't enough.

I would appreciate it if there are source codes of XDoG algorithm. Thank you!

[hasagar97]

Hi, @cating341 What XDoG implementation did you use? Also were you able to improve your results?

[Ruminum]

Hi @cating341 and @hasagar97 ,

Would you mind sharing how to run this project? I tried to run this project but always got error due to no images found even though I already put some image for training in ../datasets/lineart . I am still learning about deep learning stuff so pardon my questions. Thank you

[cating341]

@hasagar97 Thank you for your reply. I followed the XDoG pipeline: But there are usually some noises in the results or the edge lines aren't so clear. Such as the result below.

I think maybe a good parameter ε is needed in the thresholding function. I set the mean value of the image as ε. Could you tell me how did you choose the ε? Thank you.