Question

System:
    python: 3.6.8 |Anaconda custom (64-bit)| (default, Dec 29 2018, 19:04:46)  [GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)]
executable: /Users/steve/miniconda3/envs/retinanet/bin/python
   machine: Darwin-18.2.0-x86_64-i386-64bit

BLAS:
    macros: SCIPY_MKL_H=None, HAVE_CBLAS=None
  lib_dirs: /Users/steve/miniconda3/envs/retinanet/lib
cblas_libs: mkl_rt, pthread

Python deps:
       pip: 19.0.3
setuptools: 40.8.0
   sklearn: 0.20.3
     numpy: 1.16.2
     scipy: 1.2.1
    Cython: 0.29.6
    pandas: 0.24.2

我正在尝试将图像转换为补丁，然后将其转换回图像并保存。似乎有两个功能可以做到这一点。

https://scikit-learn.org/stable/modules/generated/sklearn.feature_extraction.image.extract_patches_2d.html#sklearn.feature_extraction.image.extract_patches_2d

https://scikit-learn.org/stable/modules/generated/sklearn.feature_extraction.image.reconstruct_from_patches_2d.html#sklearn.feature_extraction.image.reconstruct_from_patches_2d

使用以下代码，重建补丁似乎只是使图像“变黑”并减小大小（原始图像大小为31MB，新大小为33KB）。没有颜色，什么都没有。垃圾。

import numpy as np
from PIL import Image
from sklearn.feature_extraction import image as skl
image = Image.open('./myimage.tif')
image = np.array(image)[..., :3]
patches = skl.extract_patches_2d(image, (300, 300), max_patches = 250)
reconstruct = skl.reconstruct_from_patches_2d(patches, image.shape)
# i need this as uint8, as i'm saving it back into an image
# i tried using skimage.img_as_ubyte, but i recieve an error
reconstruct = reconstruct.astype(np.uint8)
Image.fromarray(reconstruct, 'RGB').save('./myimage.out.tif')

我在做错什么吗？从extract_patches_2d返回的补丁都具有颜色并且看起来不错。因此，我希望重建的行为相同。

重建sklearn图像补丁会产生空白图像

0 个答案: