Vectorizing all pixels as separate polygons using rasterio

Question

I use rasterio.features.shapes and use the geometry of pixels in raster data to create polygons for each pixel. Pixels with the same value merge these pixels if they are adjacent.

What I want is for each pixel to be in the form of separate polygons. How can I do it?

import rasterio
from rasterio.features import shapes
mask = None
with rasterio.Env():
    with rasterio.open(r"C:\Users\Lenovo\Desktop\clipped.tiff") as src:
        image = src.read(1) # first band
        results = (
        {'properties': {'raster_val': v}, 'geometry': s}
        for i, (s, v) 
        in enumerate(
            shapes(image, transform=src.transform)))
geoms = list(results)

The result (the red area is the combination of two same value pixels):

Expected result (separate polygons for all pixels):

Please edit the title of your question. There is no error with "shapes" because the function returns an expected result https://rasterio.readthedocs.io/en/stable/api/rasterio.features.html "Get shapes and values of connected regions in a dataset or array." What you need is some other method. — user30184, Mar 23 '23 at 15:49
A few alternatives: https://gis.stackexchange.com/questions/294592/converting-raster-to-vector-and-creating-polygons-based-on-each-pixel-in-qgis — user30184, Mar 24 '23 at 07:48
How about https://docs.dea.ga.gov.au/notebooks/Frequently_used_code/Polygonise_pixel_edges.html? Or https://spatial-dev.guru/2022/04/16/polygonize-raster-using-rioxarray-and-geopandas/? — user30184, Mar 24 '23 at 09:01

BERA · Accepted Answer · 2023-03-27T17:19:53.913

When two adjacent pixels have the same value they are dissolved into one.

You can:

add a random float value between 0-0.9999 to each pixel value to make all pixel values unique,
vectorize

then floor them back to the original values:

import rasterio
from rasterio.features import shapes
import numpy as np
import geopandas as gpd
file = r"C:\GIS\data\testdata\random_uint16.tif"
mask = None
with rasterio.Env():
with rasterio.open(file) as src:
    image = src.read(1) # first band
    #image[:2,:2]
    #array([[7, 4],
    #       [2, 7]], dtype=int16)
newimage = np.random.uniform(size=image.shape).astype(np.float32) #Create an array of random floats between 0-1 with the same shape as the raster
newimage = image+newimage #Sum the original and random array
#array([[7.9048324, 4.694755 ],
#       [2.716989 , 7.423101 ]], dtype=float32)

results = (
{'properties': {'raster_val': v}, 'geometry': s}
for i, (s, v) 
in enumerate(
    shapes(newimage, mask=mask, transform=src.transform)))


geoms = list(results)
df  = gpd.GeoDataFrame.from_features(geoms)
df = df.set_crs(3006)
df["raster_val"] = np.floor(df["raster_val"]).astype(int) 
#Restore the original raster values
df.to_file(r"C:\GIS\data\testdata\random_unint16.shp")

Vectorizing all pixels as separate polygons using rasterio

1 Answers1