scikit-image saves binary image as completely black image

问题

So, I am trying to get binary image with scikit-image and save it on disk with the following code:

gray = color.rgb2gray(img)
thresh = filters.threshold_otsu(gray)
binary = gray >= thresh
io.imsave("./testout/" + img_name, binary)

When I do io.imshow(binary), I get what I expected. But the imsave() return to me completely black image, as if it turn both True and False values into (0,0,0) in rgb or something.

So what is the right way to do it?

回答1:

from skimage import img_as_uint
# ...
io.imsave("./testout/" + img_name, img_as_uint(binary))

This seems to work, but I'm not sure it's the best way to do it.

Also, there's an issue opened on scikit-image repo: https://github.com/scikit-image/scikit-image/issues/1623

回答2:

Simply casting binary into float also does the trick:

binary = (gray >= thresh).astype(float)

回答3:

I prefer to use img_as_ubyte:

from skimage import img_as_ubyte
io.imsave("demo.jpg", img_as_uint(binary))

Besides, you don't get to see the Lossy conversion from uint16 to uint8. Losing 8 bits of resolution... warning from img_as_uint

来源：https://stackoverflow.com/questions/36206321/scikit-image-saves-binary-image-as-completely-black-image