python - 高斯滤波器后合并接近的对象

Question

我试图计算图像中的对象数量，并偶然发现了另一个问题中的以下代码，我将其应用到我的示例图像上，如下所示:

import numpy
import matplotlib.pyplot as plt
from scipy import misc, ndimage
from skimage import feature
from skimage.filters import roberts, sobel

im = misc.imread('/home/nvp/temp/kaw.png',flatten=True)
im = im.astype('int32')
edges1 = feature.canny(im, sigma=3)
plt.imshow(edges1,interpolation='nearest')
dx = ndimage.sobel(im, 1)  # horizontal derivative
dy = ndimage.sobel(im, 0)  # vertical derivative
mag = numpy.hypot(dx, dy)  # magnitude
mag *= 255.0 / numpy.max(mag)
dna = mag
dnaf = ndimage.gaussian_filter(dna, 7)
T = 27 # set threshold by hand to avoid installing `mahotas` or
       # `scipy.stsci.image` dependencies that have threshold() functions

# find connected components
labeled, nr_objects = ndimage.label(dnaf > T) # `dna[:,:,0]>T` for red-dot case
print(dnaf, labeled,len(labeled))
print("Number of objects is %d " % nr_objects)

# show labeled image
####scipy.misc.imsave('labeled_dna.png', labeled)
####scipy.misc.imshow(labeled) # black&white image
import matplotlib.pyplot as plt
plt.imsave('labeled_dna.png', labeled)
plt.imshow(labeled)
plt.show()

但是它输出以下图像:

最终的对象存储在 labeled np.array 中，我想，我想做的是合并此输出数组中的接近对象。正如您所看到的，最后一行中的第二个对象有两个部分，但它们彼此非常接近。

由于我不了解 numpy ，我想要一种设置阈值并合并对象之间距离较小的对象的方法。任何帮助表示赞赏。谢谢:)

Answer 1

您要合并的两个对象与第一行的对象 3 和 4 一样接近...因此仅基于对象之间的接近度的解决方案不会给您带来您想要的结果。

一种解决方案可能是 dilate根据物体的面积(扩大更多的小物体)。

示例(未测试):

import scipy.ndimage.morphology as morpho

# parameters
max_dilat = 20 # dilation (in number of pixels) for a small object
sz_small = 100 # size of a small object (max dilated)
sz_big   = 10000 # size of a big object (not dilated)

result = labeled*0

# for each detected object
for obj_id in range(1, nr_objects+1):
    # creates a binary image with the current object
    obj_img = (labeled==obj_id)
    # computes object's area
    area = numpy.sum(obj_img)
    # dilatation factor inversely proportional to area
    dfac = int( max_dilat*(1-min(1,(max(0,area-sz_small)/sz_big))) )
    # dilates object
    dilat = morpho.binary_dilation(obj_img, iterations=dfac)

    result += dilat

# result is now an int array with ones where you have a single 
# dilated object, and twos (or more) in overlapping parts
labeled, nr_objects = ndimage.label(result>0)