CODEWITHSUNDEEP
pythoncoordinatespixelrastergdal
niaR
niaR

Reputation: 107

pixel/array position to lat long gdal Python

I am trying to convert positions in a raster representing a .tif to the corresponding global coordinates. Converting the whole array to a tif and load it to QGIS everything is referenced fine, but using the below calculation method for single points there is a slight offset (to east-north-east in the resulting coordinates....

raster.tif uses ETRS 89 UTM Zone 32N

Does anyone has an idea?

from osgeo import ogr, gdal, osr
import numpy as np


raster = gdal.Open("rasters/raster.tif")
raster_array = np.array(raster.ReadAsArray())


def pixel2coord(x, y):
     xoff, a, b, yoff, d, e = raster.GetGeoTransform()
     xp = a * x + b * y + xoff
     yp = d * x + e * y + yoff
     return(xp, yp)


print(pixel2cood(500,598))

Upvotes: 7

Views: 6140

Answers (2)

mobiuscreek
mobiuscreek

Reputation: 421

There is no reason to do this manually. You just need to install rasterio, a python library based on GDAL and numpy. Even if the image has an offset/rotation rasterio will take care of it.

You just have to do:

import rasterio

with rasterio.open('rasters/raster.tif') as map_layer:
    coords2pixels = map_layer.index(235059.32,810006.31) #input lon,lat
    pixels2coords = map_layer.xy(500,598)  #input px, py

Both functions return a tuple of the pixels and coordinates respectively.

Upvotes: 6

t_prz
t_prz

Reputation: 138

I think the issue might be that the xoff and yoff contain coordinates of the upper left corner of the most upper left pixel, and you need to calculate coordinates of the pixel's center.

def pixel2coord(x, y):
    xoff, a, b, yoff, d, e = raster.GetGeoTransform()

    xp = a * x + b * y + a * 0.5 + b * 0.5 + xoff
    yp = d * x + e * y + d * 0.5 + e * 0.5 + yoff
    return(xp, yp)

Upvotes: 5

Related Questions