FrontDetection/FrontPostProcessing.py at master · stnie/FrontDetection · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
import numpy as np
from skimage import measure, morphology

# Returns filtered fronts excluding a border region
def filterFronts(image, border):
    threshold = 0.45
    minlen = 2
    filteredImage = np.zeros_like(image)
    numChannels = image.shape[-1]
    assert(numChannels > 1)
    for channel in range(numChannels):
        thinImg = image[:,border:-border, border:-border,channel]>threshold
        labeledImage = morphology.binary_dilation(thinImg, selem = np.ones((1,3,3)))
        #labeledImage = morphology.skeletonize(labeledImage)
        labeledImage = measure.label(labeledImage, background = 0)
        labeledImage *= thinImg
        numLabels = np.max(labeledImage)
        for pidx in range(1,numLabels+1):
            singleLabel = (labeledImage == pidx)*1
            points = np.nonzero(singleLabel)
            if len(points[0]) < minlen:
                labeledImage[points] = 0
        filteredImage[:,border:-border,border:-border,channel] = (labeledImage>0)*1
    return filteredImage

def filterFrontsFreeBorder(image, northBorder, southBorder, westBorder, eastBorder):
    threshold = 0.45
    minlen = 2
    filteredImage = np.zeros_like(image)
    numChannels = image.shape[-1]
    assert(numChannels > 1)
    for channel in range(numChannels):
        thinImg = image[:,northBorder:-southBorder, westBorder:-eastBorder,channel]>threshold
        labeledImage = morphology.binary_dilation(thinImg, selem = np.ones((1,3,3)))
        #labeledImage = morphology.skeletonize(labeledImage)
        labeledImage = measure.label(labeledImage, background = 0)
        labeledImage *= thinImg
        numLabels = np.max(labeledImage)
        for pidx in range(1,numLabels+1):
            singleLabel = (labeledImage == pidx)*1
            points = np.nonzero(singleLabel)
            if len(points[0]) < minlen:
                labeledImage[points] = 0
        filteredImage[:,northBorder:-southBorder,westBorder:-eastBorder,channel] = (labeledImage>0)*1
    return filteredImage