我怎么知道python中某些坐标(存储在列表中)的二进制图像中是否有白色像素?

时间:2013-10-27 15:48:48

标签: python image-processing numpy scikit-image mahotas

我有一个numpy array binary(黑白)图像和坐标列表中的元组,如:

coordlist =[(110, 110), (110, 111), (110, 112), (110, 113), (110, 114), (110, 115), (110, 116), (110, 117), (110, 118), (110, 119), (110, 120), (100, 110), (101, 111), (102, 112), (103, 113), (104, 114), (105, 115), (106, 116), (107, 117), (108, 118), (109, 119), (110, 120)]

或作为:

coordx = [110, 110, 110, 110, 110, 110, 110, 110, 110, 110, 110, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110]
coordy = [110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120]

如何使用该坐标列表检查图像中是否存在“白色”像素? 我还想检查距离该坐标列表大约3个像素范围的白色像素。

即:

for i, j in coordx, coordy:
    for k in a range (k-3, k + 3)
        for l in a range (l-3, l + 3)
            #checking white pixels also for pixel near coordinates list

我想到了“哪里”的功能。

from skimage import morphology
import numpy as np

path = 'image/a.jpg'
col = mh.imread(path)
bn0 = col[:,:,0]
bn = (bn0 < 127)
bnsk = morphology.skeletonize(bn)
bnskInt = np.array(bnsk, dtype=np.uint8)

#finding if there are white pixel in the coord list and around that in a 5 pixel range
for i in coordlist:
np.where(?)

更新

我尝试使用shape(128,128)而不是(128,128,3),因为我的图像有这样的形状:(a,b)但是现在它找不到白色像素! 为什么以这种方式找到了什么?

    white_pixel = np.array([255, 255])
    img = np.random.randint(0, 256, (128, 128))
    print(img[150])
    print(img.shape)
    img[110, 110] = 255
    img[109, 110] = 255

    mask = np.zeros((128, 128), dtype=bool)
    mask[coordx, coordy] = 1
    #structure = np.ones((3, 3, 1))
    #mask = scipy.ndimage.morphology.binary_dilation(mask, structure)

    is_white = np.all((img * mask) == white_pixel, axis=-1)

    # This will tell you which pixels are white
    print np.where(is_white)

    # This will tell you if any pixels are white
    print np.any(is_white)

输出:

(array([], dtype=int32),)
False

1 个答案:

答案 0 :(得分:4)

更新,我已经更新了使用二进制或灰度图像的答案。请注意,图像强度现在只是标量而不是(R,G,B)值,所有图像,蒙版和结构元素都是2d阵列而不是3d阵列。您可能需要调整white_pixel的值(或以其他方式修改此代码以满足您的需要)。

import numpy as np
from skimage.morphology import binary_dilation
# Setup
coordx = [110, 110, 110, 110, 110, 110, 110, 110, 110, 110, 110, 100, 101, 102,
          103, 104, 105, 106, 107, 108, 109, 110]
coordy = [110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 110, 111, 112,
          113, 114, 115, 116, 117, 118, 119, 120]
img = np.random.random((128, 128))
img[110, 110] = 1.
img[109, 110] = 1.


# values grater than white_pixel will get detected as white pixels
white_pixel = 1

mask = np.zeros((128, 128), dtype=bool)
mask[coordx, coordy] = 1

structure = np.ones((7, 7))
mask = binary_dilation(mask, structure)

is_white = (img * mask) >= white_pixel

# This will tell you which pixels are white
print np.where(is_white)

# This will tell you if any pixels are white
print np.any(is_white)

原始答案:

如果你想知道哪些像素为白色,你只需要使用numpy.where。我只是将图像乘以掩码并使用np.any,如下所示:

# Setup
coordx = [110, 110, 110, 110, 110, 110, 110, 110, 110, 110, 110, 100, 101, 102,
          103, 104, 105, 106, 107, 108, 109, 110]
coordy = [110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 110, 111, 112,
          113, 114, 115, 116, 117, 118, 119, 120]
white_pixel = np.array([255, 255, 255])
img = np.random.randint(0, 256, (128, 128, 3))
img[110, 110, :] = 255
img[109, 110, :] = 255

mask = np.zeros((128, 128, 1), dtype=bool)
mask[coordx, coordy] = 1

structure = np.ones((7, 7, 1))
mask = binary_dilation(mask, structure)

is_white = np.all((img * mask) == white_pixel, axis=-1)

# This will tell you which pixels are white
print np.where(is_white)

# This will tell you if any pixels are white
print np.any(is_white)