async-tiff 0.5.0

No project description provided

async-tiff

Fast, low-level async TIFF reader for Python.

This documentation is for the Python bindings. Refer here for the Rust crate documentation.

For a higher-level API to read GeoTIFF files, visit async-geotiff.

Examples

Reading NAIP

from async_tiff import TIFF
from async_tiff.store import S3Store

# You'll also need to provide credentials to access a requester pays bucket
store = S3Store("naip-visualization", region="us-west-2", request_payer=True)
path = "ny/2022/60cm/rgb/40073/m_4007307_sw_18_060_20220803.tif"

tiff = await TIFF.open(path, store=store)
primary_ifd = tiff.ifds[0]

primary_ifd.geo_key_directory.citation
# 'NAD83 / UTM zone 18N'

primary_ifd.geo_key_directory.projected_type
# 26918
# (EPSG code)

primary_ifd.sample_format
# [<SampleFormat.Uint: 1>, <SampleFormat.Uint: 1>, <SampleFormat.Uint: 1>]

primary_ifd.bits_per_sample
# [8, 8, 8]

tile = await tiff.fetch_tile(0, 0, 4)
array = await tile.decode()

# Use rasterio and matplotlib for visualization
import numpy as np
from rasterio.plot import reshape_as_raster, show

# Zero-copy conversion of the rust array into a numpy array
np_array = np.asarray(array)

# Then we need to reshape the "image" axes into "raster" axes
# https://rasterio.readthedocs.io/en/stable/topics/image_processing.html
show(reshape_as_raster(np_array), adjust=True)

Reading Sentinel 2 L2A

import numpy as np
from async_tiff import TIFF
from async_tiff.store import S3Store

store = S3Store("sentinel-cogs", region="us-west-2", skip_signature=True)
path = "sentinel-s2-l2a-cogs/12/S/UF/2022/6/S2B_12SUF_20220609_0_L2A/B04.tif"

tiff = await TIFF.open(path, store=store)
primary_ifd = tiff.ifds[0]
# Text readable citation
primary_ifd.geo_key_directory.citation
# EPSG code
primary_ifd.geo_key_directory.projected_type

primary_ifd.sample_format[0]
# <SampleFormat.Uint: 1>
primary_ifd.bits_per_sample[0]
# 16

tile = await tiff.fetch_tile(0, 0, 0)
array = await tile.decode()

# Zero-copy conversion of the rust array into a numpy array
np_array = np.asarray(array)
np_array.shape
# (1024, 1024, 1)