DISCLAMER:

This software was developed in the Vision Laboratory at York University Department of 
Computer Science and Engineering and the Centre for Vision Research by Mikhail Sizintsev.
This is experimental software. Use at your own risk. No warranty is implied by this
distribution. The code is designed for research purpose only and cannot be used
for commercial purposes.

The following package contains a C++ implementation of 
generalized integral images (GII) [Derpanis, Leung and Sizintsev'07]
and SIFT detector that uses GII for testing purposes.

The project has been tested under VC++.Net 2005
To complile the project, include all files there.

File list:

GeneralIntegralImage.cpp
Image (.h/.cpp)
IntegralImage  (.h/.cpp)
ScaleSpace.h
IntegralSift (.h/.cpp)

IntegralImage (.h/.cpp) and Image are (.h/.cpp) the main and only files to handle GII.

IntegralSift(h/cpp) and ScaleSpace.h are files to that implement B-spline scale-space 
and SIFT detector (DoB as referred in the paper). The detector output is stored in 
file tmp.key in the format as in original [Lowe'04] paper.

	Usage:
		Executable.exe 'image.pgm' order_of_GII max_T
	Example run:
		GeneralIntegralImage.exe cat.pgm 2 15
The program Will detect sift features using second-order integral imageaand record 
detected features (without descriptors) into tmp.key file. The Scales-space is
be constructed using first-order B-splines with width (referred as T in the paper) 
iterating from 1 to 15 in this particular example.


References:

[Derpanis, Leung and Sizintsev'07] K.G. Derpanis, E.T.H Leung and M. Sizintsev 
	   "Fast Scale-Space Feature Representations by Generalized Integral Images",
           International Conference on Image Processing, 2007.

[Lowe'04] D.G. Lowe "Distinctive image features from scale-invariant keypoints", 
International Journal of Computer Vision 2004.
