Python `filter.sobel()`输入

Question

尝试使用Python语言中的filter.sobel()函数。

from PIL import Image
from skimage import data, io, filter
#...

image = Image.open('ME.jpg')
#right below is what it would normally be. I tried to substitute the code above for this
#image = data.coins()
edges = filter.sobel(image)
io.imshow(image)
io.show()
io.imshow(edges)
io.show()

o/p：

Traceback (most recent call last):
 File "edg.py", line 7, in <module>
    edges = filter.sobel(image)
  File "/Library/Python/2.7/site-packages/skimage/filter/edges.py", line 83, in sobel
    return np.sqrt(hsobel(image, mask)**2 + vsobel(image, mask)**2)
  File "/Library/Python/2.7/site-packages/skimage/filter/edges.py", line 114, in hsobel
    result = np.abs(convolve(image, HSOBEL_WEIGHTS))
  File     "/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/scipy/ndimage/filt    ers.py", line 664, in convolve
    origin, True)
  File     "/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/scipy/ndimage/filt    ers.py", line 498, in _correlate_or_convolve
    raise RuntimeError('filter weights array has incorrect shape.')
RuntimeError: filter weights array has incorrect shape.

如何将图像转换为类似于data.coins()中的数组

Answer 1

你为什么不使用scikit-image的图片阅读器呢？否则，只需使用黑白版本来运行算法。为此，只需将参数'as_grey‘传递给imread (我正在使用scikit image)：

image = skimage.io.imread(image_path,as_grey=True)

此参考可能会对您有所帮助：

http://scikit-image.org/docs/dev/api/skimage.io.html#imread