How to open a TIFF file created by Earth Engine with matplotlib

Question

I'm using the Python API of GEE and I try to display my results in a PDF file. I'd like to create it with matplotlib so that I can check my figure before exportation. To add a little context :

I create an image using :

task_config = {
    'image':clip,
    'description':description,
    'scale': 30,
    'region':buffers[index],
    'maxPixels': 1e12
}
    
task = ee.batch.Export.image.toDrive(**task_config)
task.start()

I can display this image in QGIS:

Then I download it to my local folder and try to read it with matplotlib:

img = plt.imread(file)
plt.imshow(img)
plt.show()

I get the following error :

UnidentifiedImageError: cannot identify image file '/home/prambaud/time_series_results/tmp/test_pts_pt_0_Red_Green_Blue_2005.tif'

Is it normal? Is there another way to open Images that are generated by GEE ?

Answer 1

It requires a bit more manipulation but in the end it worked :

import rasterio as rio
import numpy as np
from matplotlib import pyplot as plt

with rio.open(file) as f:
    data = f.read()
                
    bands = [] 
    for i in range(3):
        band = data[i]
        #remove the NaN from the analysis
        # 3000 is the unit for landsat data change it according to your need
        h_, bin_ = np.histogram(band[np.isfinite(band)].flatten(), 3000, density=True) 
    
        cdf = h_.cumsum() # cumulative distribution function
        cdf = 3000 * cdf / cdf[-1] # normalize
    
        # use linear interpolation of cdf to find new pixel values
        band_equalized = np.interp(band.flatten(), bin_[:-1], cdf)
        band_equalized = band_equalized.reshape(band.shape)
        
        bands.append(band_equalized)
    
    data = np.stack( bands, axis=0 )

    data = data/3000
    data = data.clip(0, 1)
    data = np.transpose(data,[1,2,0])
    plt.imshow(data, interpolation='nearest')
    plt.show()