A simple example of image matcher using KAZE local feature which is published in ECCV2012.

Pablo F. Alcantarilla, Adrien Bartoli and Andrew J. Davison., "KAZE Features," In European Conference on Computer Vision (ECCV), Fiorenze, Italy, October 2012. (Project Page, Implementation)

And this sample code is developed by @belltailjp personally for the blog article ECCV2012で発表されたKAZE局所特徴量を試してみた | さかな前線 (Japanese).

This sample code is confirmed to work on the following configuration:

• Linux Mint 3.6 (Equivalent to Ubuntu 12.04)
• GCC 4.7
• OpenCV 2.4.3
• Boost C++ Library
• KAZE implementation in the same directory as the sample code

And a webcamera is needed for demo.cpp

# To build simulation matcher
% g++ simulation.cpp KAZE.cpp Ipoint.cpp nldiffusion_functions.cpp utils.cpp \
    -lopencv_core -lopencv_highgui -lopencv_imgproc -lopencv_legacy -lopencv_features2d -lopencv_nonfree \
    -lboost_system -lboost_thread -O3
% ./a.out

# To build the intractive matcher demo
% g++ demo.cpp KAZE.cpp Ipoint.cpp nldiffusion_functions.cpp utils.cpp \
    -lopencv_core -lopencv_highgui -lopencv_nonfree -lopencv_imgproc -lopencv_legacy -lopencv_features2d \
    -lboost_system -lboost_thread -O3
% ./a.out

Currently there have been several improvements introduced to KAZE (like Accelerated-KAZE), and it is already implemented in OpenCV 3.0 natively. Please be noted that my sample code will no longer be maintained and supported. Refer AKAZE local features matching — OpenCV 3.0.0-dev documentation for more details.