skimage/imageio读取灰度PNG图片失败的问题求助

此前长期使用skimage.io.imread或imageio.imread读取灰度图片正常，未修改代码的情况下突然读取失败，但合作者可正常读取目标文件。读取彩色图片无异常，已重装scikit-image、imageio、Pillow库仍无法解决。

图片属性信息 #

使用ImageMagick的identify命令查看图片属性：

identify test/resources/biosynth1_cropped/text_removed.png 
test/resources/biosynth1_cropped/text_removed.png PNG 1512x315 1512x315+0+0 8-bit sRGB 48c 10094B 0.000u 0:00.005

添加--verbose参数查看的部分属性：

Type: Grayscale
  Base type: Undefined
  Endianness: Undefined
  Depth: 8-bit
  Channel depth:
    Red: 8-bit
    Green: 8-bit
    Blue: 8-bit
  Channel statistics:
    Pixels: 476280

测试代码 #

from skimage import io
import imageio
file="/Users/pm286/workspace/pyamiimage/test/resources/biosynth1_cropped/arrows_removed.png" 
img = imageio.imread(file)

报错信息 #

(base) pm286macbook:pyamiimage pm286$ python test_load.py 
Traceback (most recent call last):
  File "test_load.py", line 9, in <module>
    img = imageio.imread(file)
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/core/functions.py", line 265, in imread
    reader = read(uri, format, "i", **kwargs)
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/core/functions.py", line 186, in get_reader
    return format.get_reader(request)
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/core/format.py", line 170, in get_reader
    return self.Reader(self, request)
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/core/format.py", line 221, in __init__
    self._open(**self.request.kwargs.copy())
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/plugins/pillow.py", line 298, in _open
    return PillowFormat.Reader._open(self, pilmode=pilmode, as_gray=as_gray)
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/plugins/pillow.py", line 138, in _open
    as_gray=as_gray, is_gray=_palette_is_grayscale(self._im)
  File "/opt/anaconda3/lib/python3.8/site-packages/imageio/plugins/pillow.py", line 689, in _palette_is_grayscale
    palette = np.asarray(pil_image.getpalette()).reshape((256, 3))
ValueError: cannot reshape array of size 144 into shape (256,3)

解决方案建议 #

• 强制指定灰度读取模式：在imread中添加as_gray=True参数，绕过imageio的调色板检测逻辑，直接生成单通道(x,y)维度的灰度数组，同时解决维度变为(x,y,3)的问题：

  # 使用imageio
  img = imageio.imread(file, as_gray=True)
  # 使用skimage
  img = io.imread(file, as_gray=True)
  

• 手动处理调色板图片：直接用Pillow打开图片，手动转换为灰度模式后再转为numpy数组，避免依赖imageio的自动解析：

  from PIL import Image
  import numpy as np
  
  img_pil = Image.open(file)
  # 若为调色板模式，转为单通道灰度
  if img_pil.mode == 'P':
      img_pil = img_pil.convert('L')
  img = np.array(img_pil)
  

• 对齐Pillow版本：检查你和合作者的Pillow版本是否一致，不同版本对调色板PNG的处理逻辑可能存在差异：

  import PIL
  print(PIL.__version__)
  

  若版本不同，尝试升级/降级到和合作者相同的版本。

• 重新保存图片为标准灰度格式：用ImageMagick或Pillow将异常图片转换为标准单通道灰度PNG，从源头上解决格式问题：

  # ImageMagick命令
  convert test/resources/biosynth1_cropped/text_removed.png -type Grayscale -depth 8 fixed_gray.png
  

  或用Python代码转换：

  from PIL import Image
  img = Image.open(file)
  img.convert('L').save('fixed_gray.png')
  

内容的提问来源于stack exchange，提问作者peter.murray.rust