Spectral Python 0.16.0 adds Adaptive Coherence Estimator (ACE) and Pixel Purity Index (PPI)

Thomas Boggs — Sun, 19 Oct 2014 16:36:47 -0000

Spectral Python (SPy) is a python package for hyperspectral image processing.

Release 0.16.0 adds Python 3 support for all functions except view_cube and view_nd. Note that for Python 3, you should use the Qt4Agg matplotlib backend.

New features in this release include the Adaptive Coherence/Cosine Esimator (ace) target detector, Pixel Purity Index (ppi), ability to save ENVI classification files (envi.save_classification), and linear contrast enhancement (by data limits or cumulative histogram percentiles). The SPy imshow function now applies a 2% histogram color stretch by default (this can be overridden in the spectral.settings object).

Additional info is in the version 0.16.0 issues.

Spectral Python 0.15.0 adds Minimum Noise Fraction (MNF)

Thomas Boggs — Thu, 05 Jun 2014 02:10:02 -0000

New Features

Added Minimum Noise Fraction (mnf) algorithm (a.k.a., Noise-Adjusted
Principal Components). An associated noise_from_diffs function enables
estimation of image noise from a homogeneous region of the image.

Changes

When calling envi.save_image, assume an ndarray with two dimensions is
a single-band image (i.e., don't require an explicit third dimension).
[Issue #9] All SpyFile subclass read methods now have an optional
use_memmap argument to indicate whether the memmap interface should be
used (vice direct file read) on a per-call basis. Default values are
specific to the particular method and file interleave.

Bug Fixes

[Issue #7] Handle recognize comment lines in ENVI headers and accept blank
parameter values in the header. Thanks to Don March (http://ohspite.net)
[Issue #2] Garbage results were being generated for several algorithms when
a NaN value was present in the image data. Reasonable checks are now
performed in several algorithms and an optional allow_nan argument (False
by default) was added to calc_stats.
[Issue #1] For images with more rows than columns, the row/col of the pixel
under the mouse cursor did not display if the row index was greater than
the image width.

Performance Improvements

[Issue #5] Improved BilFile.read_bands performance. Thanks to Don March
(http://ohspite.net)
[Issue #8] Faster creation/display of RGB images for display. Thanks to
Don March (http://ohspite.net)

New GUI Features and Performance Improvements in Spectral Python (SPy) 0.13

Thomas Boggs — Mon, 06 Jan 2014 01:54:08 -0000

The SPy imshow wrapper around matplotlib's imshow function provides numerous new features, including:

Interactive image class labeling using keyboard & mouse
Zoom windows
Class overlays with adjustable transparency
Dynamic view of changing pixel classes when modified in an ND Window.

Data/Statistic cacheing and more efficient use of numpy provides significant performance improvement in mutiple algorithms (GMLC 14x, Mahalanobis classifier 8x, kmeans 3x). Functions rx and matched_filter are significantly faster, particularly when using common global covariance.

The new cov_avg function computes covariance averaged over a set of classes (useful when samples are limited or global covariance is desired). Christian Mielke provided code for the msam function, which computes the Modified SAM score (by Oshigami et al).

Spectral Python 0.12 has been released

Thomas Boggs — Sat, 07 Sep 2013 02:13:08 -0000

SPy 0.12 adds the ability to use local image statistics for RX anomaly detection. A new memmap interface enables simplified creation of synthetic hyperspectral images. The ability to suppress progress status messages and the addition of a wrapper around matplotlib's imshow function enable simplified integration of SPy code in IPython Notebooks.

SPy 0.11 adds anomaly detection and ability to save/create hyperspectral image files.

Thomas Boggs — Thu, 04 Apr 2013 01:39:40 -0000

Release 0.11 of Spectral Python (SPy) adds an RX anomaly detector, ability to save and create images in ENVI format (see envi.save_image and envi.create_image), and a unit-testing sub-package. The top-level namespace has been simplified and several functions have been renamed for consistency (image is now open_image and save_image is now save_rgb).

SPy 0.10.1 bug-fix release is now available

Thomas Boggs — Sat, 23 Feb 2013 05:18:15 -0000

This is a bug-fix release that corrects the spectrum displayed when double-
clicking on a raster display. Version 0.10 introduced a bug that had the
row/column swapped, resulting in either the wrong pixel being plotted or an
exception raised.

If you have installed SPy 0.10, you should install this update as soon as possible.

Spectral Python now uses IPython

Thomas Boggs — Sun, 17 Feb 2013 19:39:06 -0000

With the release of Spectral Python (SPy) 0.10, SPy now uses IPython for non-blocking GUI functions (raster views, hypercubes, N-Dimensional scatter plots, and spectral plots). To enable this capability, start IPython in pylab mode before importing spectral (see the SPy User's Guide for details).

Spectral Python 0.8 Released

Thomas Boggs — Thu, 24 Jan 2013 00:25:04 -0000

SPy 0.8 adds a linear matched filter target detector (MatchedFilter) and provides a generic LinearTransform object that can be applied to a SpyFile object or numpy.ndarray.

Spectral Python N-Dimensional Visualization

Thomas Boggs — Sun, 15 Jul 2012 21:48:58 -0000

Spectral Python (SPy) 0.8 adds an N-dimensional visualization capability for hyperspectral image data. The N-D display allows up to 24 features to be displayed simultaneously in an interactive, 3D display. Data points (HSI pixels) can be selected in the display to identify their location in the source image and regions of pixels can be selected to reassign their class IDs. See the updated documentation for details (http://spectralpython.sourceforge.net/graphics.html#nd-displays).

Spectral Python (SPy) 0.7 is now available

Thomas Boggs — Tue, 28 Feb 2012 18:56:04 -0000

Reading sections of images is about 10 times faster now. The k-means algorithm is also about 10 times faster (for in-memory images). Numerous methods/functions have been renamed for consistency with external packages.

Recent posts to news