First release in ages!

• Processing 3 compatibility via fixing size() calls in all the examples (thanks @nextgensparx!)
• Raspberry Pi support through armv6hf compatibility (thanks @gohai!)
• Wrappers for some morphological operations (thanks @qwzybug!)

• Implementation of Optical Flow using the Farneback algorithm. See the example
• Support for arm7 devices (Thanks, @jmwohl!)

• ByteBuffer loop-less PImage conversion technique from @mudphone should cause significant speedup
• Updated all examples to not explicitly call P2D in setup() since that seems to cause problems with contemporary processing.

• Add a version of detect() that allows additional options for detection speed (thanks @eyaler).
• Add a version of loadCascade() that accepts an absolute path to the cascade xml file so you can load your own.
• Add a flip() function for @bengrosser.
• Add a helper function for converting OpenCV Rect[] to java.awt Rectangle[] so people can experiment with other detectMultiScale() options.
• Add significant amounts of additional Javadocs.

• Finally, actually, fixing support for 64-bit Windows, courtesy of @zmike808!

Note: Despite what Github might say, this release was created Mon Jan 20 23:51:58 2014 -0500

• Include binaries for 64-bit windows
• Fixes from @P0ulp for export under Processing 2.1

Features:

• Official support for 32- and 64-bit Linux (thanks Arturo Castro!)
• ConvexHull support for Contours
• First release meant for inclusion in the Processing contributions manager.

Much-improved Windows support and Eclipse compatibility thanks to @P0ulp! Improvements. Plus a few other things!

• Cascade paths calculate correctly on Windows so face detection works.
• Histograms draw correctly on Windows
• Updated libs structure
• Improved Eclipse compatibility
• blur() has options for directionality, i.e. horizontal- or vertical-only blurring

First real public release! Features include:

• Load PImages and files into OpenCV
• Support for a bunch of filters: blur, brightness/contrast, threshold, adaptive threshold, range, invert, dilate and erode, histogram equalization, diff, Canny, Sobel, and Scharr edges
• Face detection and other cascade-based detection
• Contour detection with polygon approximation
• Ability to work in grayscale, RGB, or HSV color space (with support for individual color channels)
• Histogram calculation
• Region of Interest support
• Basic camera calibration support
• Background subtraction
• Min and max support
• Based on OpenCV 2.4.5
• Tested on 64-bit Mac OS X with Mountain Lion and Lion as well as 32-bit Windows 7

This could be it! Finished the last of the features on the 0.4 milestone. Need to get some additional testing, but this could be the 0.4 release version. Changes:

• renamed Mat variables with mat-prefixed instead of buffer-prefixed names
• renamed Mat access functions to not include the term "buffer"