Show HN: Select Sections from Images of Newspaper Clippings Using OpenCV, Python

38 points by code_devil 5 years ago | 7 comments
  • nurettin 5 years ago
    Striking similarity to this article, with a lot of the preprocessing steps missing.

    https://medium.com/@neshpatel/solving-sudoku-part-ii-9a7019d...

    • udayrddy 5 years ago
      So,you view the image using CV and write down the coordinates of the puzzle, put them in array and use CV to save the puzzle with different name.

      Basically, you are cropping the image and saving it - which even a crappy cropping tool does. Do not consider this as a discouraging comment - programming can do better.

      I understand you are looking for jobs and trying things. If I'm the hiring manager, I would rather prefer a candidate who found puzzle coordinates from the image bits/pixels and suggest cropping regions.

      • code_devil 5 years ago
        Actually I am currently not looking as I am backpacking :)

        I am just exploring the OpenCV library for fun things to do and in the process documenting simpler and smaller stuff for my reference. I do plan to have a follow up post that will automatically find the corners!

      • johnchristopher 5 years ago
        Nice. Would you do a follow-up showing how to automatically select corners like office lens or Dropbox do ? (I find the office lens app really really good)
        • code_devil 5 years ago
          Yes, Indeed that’s the direction I am heading towards, but I am initially starting with simple posts to keep the flow ( for myself) and also easier for people to notice the simplicity of openCV.
        • 5 years ago
          • amelius 5 years ago
            I would have liked to see the intermediate results in the article as I'm on mobile now and can't easily run the code.
            • code_devil 5 years ago
              I will update it with the intermediate step!
            • cogitotech 5 years ago
              https://cogitotech.blogspot.com/p/image-annotation-service.h...