Euclidean distance between two images



In [1]:

    
import ee
ee.Initialize()



In [2]:

    
from geetools import tools, algorithms



In [3]:

    
bands = ['B1','B2','B3']



In [4]:

    
image1 = ee.Image('LANDSAT/LC08/C01/T1_SR/LC08_232089_20170126').select(bands)



In [5]:

    
image2 = ee.Image('LANDSAT/LC08/C01/T1_SR/LC08_232089_20170211').select(bands)



In [6]:

    
image3 = ee.Image('LANDSAT/LC08/C01/T1_SR/LC08_232089_20170227').select(bands)



In [7]:

    
p = ee.Geometry.Point(-72, -42)



In [8]:

    
data1 = tools.image.getValue(image1, p, 30, side='client')



In [9]:

    
data1









    Out[9]:





{'B1': 77, 'B2': 107, 'B3': 290}



In [10]:

    
data2 = tools.image.getValue(image2, p, 30, side='client')



In [11]:

    
data2









    Out[11]:





{'B1': 90, 'B2': 112, 'B3': 272}



In [12]:

    
data3 = tools.image.getValue(image3, p, 30, side='client')



In [13]:

    
data3









    Out[13]:





{'B1': 63, 'B2': 87, 'B3': 193}



In [14]:

    
distance = algorithms.euclideanDistance(image1, image2)



In [15]:

    
data_dist = tools.image.getValue(distance, p, 30, side='client')



In [16]:

    
data_dist









    Out[16]:





{'distance': 22.759613353482084}

Compute locally to compare



In [17]:

    
import math



In [18]:

    
# Euclidean distance
def distanceF(arr1, arr2):
    zipped = zip(arr1, arr2)
    accum = 0
    for a, b in zipped:
        rest = (a-b)
        calc = rest*rest
        accum += calc
    return math.sqrt(accum)



In [19]:

    
arr1 = data1.values()
arr1









    Out[19]:





dict_values([77, 107, 290])



In [20]:

    
arr2 = data2.values()
arr2









    Out[20]:





dict_values([90, 112, 272])



In [21]:

    
local_distance = distanceF(arr1, arr2)



In [22]:

    
local_distance









    Out[22]:





22.759613353482084

Distance sum between one image and a collection of images



In [23]:

    
col = ee.ImageCollection.fromImages([image2, image3])



In [24]:

    
summed = algorithms.sumDistance(image1, col)



In [25]:

    
summed_data = tools.image.getValue(summed, p, 30, side='client')



In [26]:

    
summed_data









    Out[26]:





{'sumdist': 122.78461022926308}

Compute locally to compare results



In [27]:

    
from copy import copy
# Sum of distances
def sum_distances(values):
    results = {}
    for i, val in enumerate(values):
        val = list(val)
        cop = copy(values)
        cop = [list(a) for a in cop]
        cop.remove(val)
        dist = 0
        for r in cop:
            r = list(r)
            d = distanceF(val, r)
            dist += d
        results[i] = dist

    return results



In [28]:

    
values = [list(data1.values()), list(data2.values()), list(data3.values())]



In [29]:

    
values









    Out[29]:





[[77, 107, 290], [90, 112, 272], [63, 87, 193]]



In [30]:

    
sum_distances(values)









    Out[30]:





{0: 122.78461022926308, 1: 109.90891053940321, 2: 187.17429406170214}

Handling mask values



In [31]:

    
masked_image = image2.updateMask(image2.gte(0).Not())



In [32]:

    
data_masked = tools.image.getValue(masked_image, p, 30, side='client')
data_masked









    Out[32]:





{'B1': None, 'B2': None, 'B3': None}



In [33]:

    
data1









    Out[33]:





{'B1': 77, 'B2': 107, 'B3': 290}



In [34]:

    
distance_masked = algorithms.euclideanDistance(image1, masked_image)



In [35]:

    
data_dist_masked = tools.image.getValue(distance_masked, p, 30, side='client')
data_dist_masked









    Out[35]:





{'distance': 318.55611750522075}



In [36]:

    
distanceF(data1.values(), [0, 0, 0])









    Out[36]:





318.55611750522075



In [37]:

    
outside = image1
inside = masked_image



In [38]:

    
col_masked = ee.ImageCollection.fromImages([inside])



In [39]:

    
summed_masked = algorithms.sumDistance(outside, col_masked)



In [40]:

    
summed_data_masked = tools.image.getValue(summed_masked, p, 30, side='client')



In [41]:

    
summed_data_masked









    Out[41]:





{'sumdist': 318.55611750522075}



In [ ]: