获取光栅图像中所有像素的像素坐标

Question

I have a raster image of shape 9000x10000 that has RGB bands. I use the below code to get the XY coordinates of all pixels in the image. But it is very slow. Is there a faster way to do it?

filename='file.dat'
inDs = gdal.Open(filename)
outDs = gdal.Translate('{}.xyz'.format(filename), inDs, format='XYZ', creationOptions=["ADD_HEADER_LINE=YES"])

I want to save the XY coordinates and the pixel values in a dataframe.

Answer 1

If your raster file has a GeoTransform attribute, you can try this:

import gdal
import pandas as pd

def ix2xy(r,c,gt):
    '''Gets x,y from row and column'''
    x = gt[0] + r * gt[1]
    y = gt[3] + c * gt[5]
    return(x,y)

This little function gets the X/Y coordinates from the GeoTransform attribute which is a tuple with (xorigin, xres, 0, yorigin, 0, yres) .

ds = gdal.Open('file.dat')
gt = ds.GetGeoTransform()


df = pd.DataFrame.from_records(itertools.product(range(ds.RasterYSize),range(ds.RasterXSize)),columns=['Row','Column'])

ds = None

df['X'], df['Y'] = zip(*df.apply(lambda x: ix2xy(x['Column'],x['Row'],gt),axis=1))

This should give you a tidy dataframe with the columns Row , Column , X and Y .