Cumulative count cut min/max values in Python different from min/max values in QGIS

Question

I am trying to replicate using Python, what QGIS does to visualize the rasters in the map.

QGIS automatically chooses as contrast enhancement the method Stretch to MinMax. Looking at the Min/Max Settings, the Cumulative count cut is selected with the range being between 2% - 98%.

I saw in this answer that the cumulative count cut method is basically the same than numpy.nanpercentile. So I implemented this function in python:

def cumulative_count_cut(band, min_percentile=2, max_percentile=98):
    min_val = np.nanpercentile(band, min_percentile)
    max_val = np.nanpercentile(band, max_percentile)
    print(min_val,max_val)

I was expecting to the Min and Max values equal or even near the ones in QGIS but this is what I got instead:

• Band 04: Min: 0.0 | Max: 1221.0
• Band 03: Min: 0.0 | Max: 1430.0
• Band 02: Min: 0.0 | Max: 1663.0

Am I missing something? Shouldn't the values be the same?

Answer 1

I have found the answer and it was quite easy.Considering that the min value was always 0, I tried to see if by ignoring the 0 values, the min_val would change. And it did! They are not exactly the same, but they are close enough.

Python:

• Band 04: Min: 161 | Max: 1958
• Band 03: Min: 179 | Max: 1714
• Band 02: Min: 122 | Max: 1476

QGIS:

Code:

def cumulative_count_cut(band, min_percentile=2, max_percentile=98):
    """Apply contrast enhancement stretch similar to QGIS"""
    new_band = band[band != 0]
    min_val = np.nanpercentile(new_band, min_percentile)
    max_val = np.nanpercentile(new_band, max_percentile)
    print(min_val,max_val)
    return (band - min_val) / (max_val - min_val + offset) * 255