In this book, a new high performance face
recognition system based on matching the colour pixel
statistics is introduced. A pre-processing phase is
introduced and applied to segment faces from the background.
Furthermore, a dedicated image equalization method is
introduced and implemented to minimize the illumination
problems of the images for further processing. The histogram
of the segmented face image as pixel statistics feature is
used for face recognition by cross correlating the histogram
of a given face and the histograms of faces in the database.
Alternatively the probability distribution functions of the
images in different colour channels, together with the
Kullback-Leibler Divergence/Distance (KLD) metric is also
used for the recognition of faces. Majority voting (MV) and
feature vector fusion (FVF) methods is briefly introduced
and applied to combine feature vectors obtained from
different colour channels in HSI and YCbCr colour spaces to
improve recognition performance.