Python-如何使用PIL分割图像

Question

I would like to know what is the best way to split a given image into multiple images where each image includes one object. For example, in vase the image includes - Tree, dog, cat, ball The output should be four different images of Tree, dog, cat, ball

BTW, I'm using PIL module in python.

Thanks

Answer 1

In short: you can't do that.

Longer answer: this apparently simple task involves not only what would be some somewhat complicated code (15-50 lines), just to separate your original image into arbitrarily shaped parts, paste them on a default background, choosing an appropriate name and saving each part.

It involves object recognizing - just like humans do - so, it is a bit strong to say "you can't do that", but that involves AI techniques for image recognition, a lot , and I mean a lot of work in getting your neural-networking tool correctly coupled and trained to identify distinguished objects. (A lot like: years with a team, not hours for a person).

And anyway, even if you'd know where to start from, PIL alone does not fit that - PIL is a quick image manipulation library with a handfull of image methods. Although most of the machinery for Python AI image processing can use PIL to convert an image to a NumPy array representation were the "real algorithms" will be run.

Now, if you don't care that the program knows which objects are in the image, just that distinct objects on the image are saved to different files, one could come with a series of image processing filters to segment the image without resourcing to AI at all - that would be just a lot of work like 3 days up to two weeks for one with some experience in image processing. A couple hours if the person is with her skills really sharp and the images have a good contrast with the background and no object superposition.