SpFeas

SpFeas is a Python library for processing spatial (contextual) image features from satellite imagery.

SpFeas has been tested on Python 2.7 and 3.x. It is dependent on MpGlue, which is used for image I/O and land cover classification.

Current version

0.3.4

Refer to the Wiki for changes.

All comments and suggestions for improvement are welcome. Please post to the issues page.

Installation

Use either Option A (Recommended) or Option B below to install Spfeas.

Option A: Installing or upgrading from source

git clone https://github.com/jgrss/spfeas.git

cd <location where you saved SpFeas>/spfeas/
git pull origin master

python setup.py build && python setup.py install

SpFeas should now be installed in the Python /site-packages directory.

Option B: Installing with the bash or CMD installers

The installers are intended to streamline the installation process, particularly if you are installing SpFeas for the first time. Please report bugs here.

1) Open INSTALLATION.ipynb under /notebooks.

2) Follow the instructions to install SpFeas for your operating system.

Checking the installation

On OSX or Linux, the following line should print something like /usr/local/bin/spfeas:

which spfeas

On Windows, the following line should print Python path\Scripts\spfeas:

where spfeas

Testing the installation

In a Python interpreter:

>>> import spfeas
>>> spfeas.test_features()

In a terminal or CMD window:

python -c "import spfeas;spfeas.test_features()"

or with Python 3.x

python3 -c "import spfeas;spfeas.test_features()"

You should see SpFeas tests were OK. if SpFeas ran as expected.

Updating

Navigate to the cloned SpFeas git repository and pull the latest version.

cd spfeas/
git pull origin master
python setup.py build && python setup.py install

Uninstall

To uninstall SpFeas, type the following line in the terminal:
pip uninstall spfeas

Usage examples

Python usage:

>>> import spfeas
>>>
>>> # Extract the local mean at scales 16x16 and 32x32, and
>>> #   save the results to 8x8 pixel blocks.
>>> spfeas.spatial_features('/input_image.tif',
>>>                         '/output_dir',
>>>                         block=8,
>>>                         scales=[16, 32],
>>>                         triggers=['mean'])

Command-line usage:

spfeas -h

spfeas -e
Traditional single pixel moving window

The block size must equal 1 and the scales must be odd and > the block size.

spfeas -i /input_image.tif -o /output_directory -tr mean hog --block 1 --scales 5 7 --sect-size 1000 --n-jobs -1 --overviews
Pixel block moving window

The block size must be even and the scales must be even and >= the block size.

spfeas -i /input_image.tif -o /output_directory -tr mean hog --block 4 --scales 4 8 --sect-size 1000 --n-jobs -1 --overviews

Detailed examples

Please refer to /notebooks/examples.ipynb.

SpFeas parameters

Naming conventions

After running SpFeas, the output files will consist of tiled GeoTiffs and a YAML information file. The image processing is performed on a tile by tile basis. Therefore, the input image will be divided into multiple, smaller tiles. The YAML file is used to monitor the tiling process, and in the event of processing failure, allows a user to continue processing from the last finished tile.

See below for the naming convention of these files.

YAML status
<OUT_DIRECTORY>/<FILENAME>__BD#_BK#_SC#_TR%.yaml

Example:
out_dir/image_name__BD1_BK4_SC4-8_TRmean-hog.yaml
Tiled files
<OUT_DIRECTORY>/<FILENAME>__BD#_BK#_SC#_TR%/<FILENAME>__BD#_BK#_SC#_ST1-###_TL######.tif

Example:
out_dir/image_name__BD1_BK4_SC4-8_TRmean-hog/image_name__BD1_BK4_SC4-8_ST1-012_TL000001.tif
out_dir/image_name__BD1_BK4_SC4-8_TRmean-hog/image_name__BD1_BK4_SC4-8_ST1-012_TL000002.tif
out_dir/image_name__BD1_BK4_SC4-8_TRmean-hog/image_name__BD1_BK4_SC4-8_ST1-012_TL000003.tif

BD = input band position
BK = input block size
SC = input scales
TR = input triggers
ST = statistics
TL = tile position

Spatial features

Below is a list of references that SpFeas draws from.

Processing

Graesser, Jordan, Cheriyadat, Anil, Vatsavai, Ranga Raju, Chandola, Varun, Long, Jordan, and Bright, Eddie (2012) Image based characterization of formal and informal neighborhoods in an urban landscape. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 5(4), 1164–1176.

Differential Morphological Profiles (dmp)

mean variance

Pesaresi, Martino and Benediktsson, Jon Atli (2001) A New Approach for the Morphological Segmentation of High-Resolution Satellite Imagery. IEEE Transactions on Geoscience and Remote Sensing, 39(2).

Pesaresi, Martino et al. (2013) A Global Human Settlement Layer From Optical HR/VHR RS Data: Concept and First Results. IEEE Transactions on Geoscience and Remote Sensing, 6(5).

Two-band Enhanced Vegetation Index (evi2)

mean variance

Jiang, Zhangyan, Huete, Alfredo R, Didan, Kamel, and Miura, Tomoaki (2008) Development of a two-band enhanced vegetation index without a blue band. Remote Sensing of Environment, 112, 3833–3845.

Fourier Transform (fourier)

mean variance

Measure of local energy (power spectrum)

Gabor filter (gabor)

mean variance x n filters

Image convolution with a series of 2d Gabor filters

Edge gradient magnitude (grad)

mean variance

Edge gradient magnitude

Histogram of Oriented Gradients (hog)

max mean variance skew kurtosis

Dalal, N and Triggs, B (2005) Histograms of Oriented Gradients for Human Detection. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005, San Diego, CA, USA.

Lacunarity (lac)

lacunarity

Myint, Soe W, Mesev, Victor, and Lam, Nina (2006) Urban Textural Analysis from Remote Sensor Data: Lacunarity Measurements Based on the Differential Box Counting Method. Geographical Analysis, ISSN 0016-7363.

Local Binary Patterns (lbpm)

max mean variance skew kurtosis

Ojala, Timo, Pietikainen, Matti, and Maenpaa, Topi (2002) Multiresolution gray-scale and rotation invariant texture classification with local binary patterns. IEEE Transactions on Pattern Analysis and Machine Intelligence, 24(7), 971–987.

Line Support Regions (lsr)

line length line mean line contrast

Ünsalan, Cem and Boyer, KL (2004) Classifying Land Development in High-Resolution Satellite Imagery Using Hybrid Structural–Multispectral Features. IEEE Transactions on Geoscience and Remote Sensing, 42(12).

Ünsalan, Cem (2006) Gradient-Magnitude-Based Support Regions in Structural Land Use Classification. IEEE Geoscience and Remote Sensing Letters, 3(4).

Inverse distance weighted mean and variance (mean)

mean variance

Pixels near the center of the local scale/window are given inversely higher weights to the farthest pixel in the window.

Normalized Difference Vegetation Index (ndvi)

mean variance

Tucker, CJ (1979) Red and photographic red linear combinations for monitoring vegetation. Remote Sensing of Environment, 8, 127–150.

Built-up presence index (pantex)

min contrast

Pesaresi, Martino, Gerhardinger, Andrea, and Kayitakire, François (2008) A Robust Built-Up Area Presence Index by Anisotropic Rotation-Invariant Textural Measure. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 1(3).

Oriented FAST and Rotated BRIEF (orb)

max mean variance skew kurtosis

Rublee, Ethan, Rabaud, Vincent, Konolige, Kurt, and Bradski, Gary R (2011) ORB: An efficient alternative to SIFT or SURF. ICCV 2011, 2564-2571.

Image saliency (saliency)

mean variance

Perazzi, Federico, Krahenb, Philipp, Pritch, Yael, and Hornung, Alexander (2012) Contrast Based Filtering for Salient Region Detection. IEEE CVPR, Providence, Rhode Island, USA, 16-21.

Cheng, Ming-Ming, Mitra, Niloy J, Huang, Xiaolei, Torr, Philip HS, and Hu, Shi-Min (2015) Global Contrast based Salient Region detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 37(3), 569–582.

Structural Feature Sets (sfs)

max line length min line length mean w-mean standard deviation max ratio of orthogonal angles

Huang, Xin, Zhang, Liangpei, and Li, Pingxiang (2007) Classification and Extraction of Spatial Features in Urban Areas Using High-Resolution Multispectral Imagery. IEEE Geoscience and Remote Sensing Letters, 4(2).

‘To Read’ list

Hoberg, Thorsten, Rottensteiner, Franz, Feitosa, Raul Queiroz, and Heipke, Christian (2015) Conditional Random Fields for multitemporal and multiscale classification of optical satellite imagery. IEEE Transactions on Geoscience and Remote Sensing, 53(2).

Kenduiywo, Benson Kipkemboi, Bargiel, Damian, and Soergel, Uwe (2017) Higher order Dynamic Conditional Random Fields ensemble for crop type classification in radar images. IEEE Transactions on Geoscience and Remote Sensing, 55(8).

Volpi, Michele and Ferrari, Vittorio (2015) Semantic segmentation of urban scenes by learning local class interactions. Computer Vision Foundation.

Benedeka, Csaba, Shadaydeha, Maha, Katob, Zoltan, Sziranyi, Tamas, and Zerubia, Josiane (2015) Multilayer Markov Random Field models for change detection in optical remote sensing images. ISPRS Journal of Photogrammetry and Remote Sensing.

Wehmann, Adam and Liu, Desheng (2015) A spatial–temporal contextual Markovian kernel method for multi-temporal land cover mapping. ISPRS Journal of Photogrammetry and Remote Sensing, 107, 77–89.

Development

For questions or bugs, please submit an issue.