/** @file RELEASE_NOTES.TXT
This file gives a high level overview of recent changes.

Revision 2.4

- WARNING: VC++ build files haven't been updated in a while, so if
  you're building on MS Windows, you'll have to update them, and 
  maybe do some light porting of the new code.
- Added KD-Tree implementation (should maybe live in a different library?).
- Scaled hue of RGB -> HSV conversion so that it is in the 
  range [0.0, 1.0].
- Some updates to edge detection code.
- Added implementation of Nister's version of the Five Point Algorithm.
- Added an implementation of Grunert's Three Point Perspective
  Pose Estimation Algorithm.
- Added tree-frame version of Nister's Five Point Algorithm.
- Added ImageWarper class to handle repeated image transformations.
- Added code to do stereo rectification of calibrated cameras.
- Added arithmetic operators for PixelRGB.
- Added YIQ colorspace and conversion operators.
- Added Extended Kalman Filter implementation.
- Added "QuadMap" data structure that might reasonably be called "QuadTree"
  instead.

Revision 2.3

- This is an internal release only.  See release notes for revision 2.4.

Revision 2.2

- WARNING: VC++ build files haven't been updated in a while, so if
  you're building on MS Windows, you'll have to update them, and 
  maybe do some light porting of the new code.
- Updated build files for recent GNU autotools versions.
- Added implemetation of Longuet-Higgins's Eight Point Algorithm.
- Added dilate() morphological function.
- Slightly changed how gray images are computed from RGB images.
- Several minor updates and bugfixes.

Revision 2.1

WARNING: VC++ build files haven't been updated in a while, so if
  you're building on MS Windows, you'll have to update them, and 
  maybe do some light porting of the new code.

IMPORTANT: This library now (maybe temporarily) depends on libpng.

- Refactored SegmenterFelzenszwalb to allow slightly more
  general application.
- Added code implementing the feature association algorithm of Scott
  and Longuet-Higgins.
- Added code to read and write png images.  This code may soon be
  migrated to another library.
- Added implementation of the RANSAC algorithm.
- Updated RGB->GRAY colorspace conversion (Note that this is a
  semantic change.  The resulting images have different values than
  they used to).
- Minor bugfixes and cleanup.

IMPORTANT: This library now depends on dlrGeometry.

- New CameraIntrinsics* classes for representing camera calibration parameters.
- Added an implementation of Felzenszwalb and Huttenlocher's graph-based
  segmentation algorithm.
- Added a few new updates, colorspace conversions, etc.
- Other minor updates, comments, etc.

Revision 2.0

- Through version 1.28, dlrComputerVision was part of the larger
  dlr_libs package.  This release makes dlrComputerVision into a
  stand-alone package.  Version 2.0 of dlrComputerVision is identical
  to the version in dlr_libs 1.28.

**/