Is there a way to use Pillows “Image.convert()” on an existing variable?

问题

Hello,

maybe this question looks stupid, but I try to use Pillows Image.convert() to convert an image to grayscale. This image I have stored in a variable img because I already pre-processed it, but not with Pillow (type: numpy.ndarray). So I type:

img = Image.convert('LA')

But it does not seem to work, as it says:

AttributeError: module 'PIL.Image' has no attribute 'convert'

If I type img = Image.open("picture.jpg").convert('LA') it works, but I want to use it on a variable that already exists. I also do not want to save the preprocessed image just to open and convert it with the previous command because this is even more inefficient (in terms of speed and CPU-power). So: Is there a proper way to do this?

Thanks for the help in advance!

回答1:

Whilst you could perfectly well convert your Numpy array to a PIL Image and then convert that to greyscale and then convert back to a Numpy array like this:

PILImage = Image.fromarray(Numpyimg)
PILgrey  = PILImage.convert('L')
Numpygrey= np.array(PILgrey)

You might as well just do the ITU-R 601-2 luma transform yourself, i.e.

L = 0.299 * Red + 0.587 * Green + 0.114 * Blue

So, you would get:

Numpygrey = np.dot(Numpyimg[...,:3], [0.299, 0.587, 0.114]).astype(np.uint8)

回答2:

You can use

img = Image.fromarray(img)

to convert to a PIL Image type. From there, you should be able to use PIL's convert() function

img = img.convert('LA')

then, to access the pixel values directly you can either convert back to a numpy array

img_array = np.asarray(img)

or get pixel access to the PIL Image using

pixels = img.load()

来源：https://stackoverflow.com/questions/58526520/is-there-a-way-to-use-pillows-image-convert-on-an-existing-variable