This is the homepage of the Matlab INFace toolbox for illumination invariant face recognition. The toolbox contains Matlab implementations of some popular photometric normalization techniques commonly used in face recognition systems. The current version of the toolbox is INFace v2.1. The toolbox v2.1 is now (26.1.2012) available for download.


About the toolbox

The INFace (Illumination Normalization techniques for robust Face recognition) toolbox is a collection of Matlab functions and scripts intended to help researchers working in the filed of face recognition. The toolbox was produced as a byproduct of the research work presented here and here. It includes implementations of several state-of-the-art photometric normalization techniques as well as a number of histogram manipulation functions, which can be useful for the task of illumination invariant face recognition.

Changes from toolbox version v2.0

Inface v2.1, january 2012

- added retina modeling based normalization technique
- fixed a bug in one of the demo scripts

Inface v2.0, october 2011

- added gamma correction
- added image range adjustment
- added threshold filtering
- unified return values in case of error
- added DoG filtering
- added Triggs and Tan method
- made histogram truncation after normalization optional
- added robust postprocessing technique from TT paper
- added Weberfaces technique
- added multi-scale Weberfaces technique
- added modified anisotropic smoothing technique
- added option for retrieving the luminance function
- fixed a bug in the DCT normalization technique
- added several new demo scripts
- update help sections of most of the functions
- added a script for install validation
- updated "other" folder with new references
- added new "postprocessing" folder to toolbox hierarchy
- created new toolbox web page
- updated user manual


by Vitomir Štruc