This is the reason that our system has to yield the better
performance even in poor illumination, cheaper cameras and
even the deviation in the light of the environment. The system
has to work within the real time constraints. Hence the
procedure has to limit the parameters to the minimal possible
level by removing the unnecessary information at the first
instance. The background subtraction is performed in the HSV
color space for removing the illumination changes where the
hand is to be modeled for simplifying the input. Thus focusing
on areas matching to human skin color. As the initialization
phase is over, the haar cascade [18] classifier is responsible
for locating hand position and classifying gestures (open,
close, pointing, etc.). Haar-Like Features [19] is features that
digitalize images to analyze images in object recognition
applications. The tracking of hand is done through camshift
technique [20] with shifting the region of interest with average
shift in the object of interest i.e. hands. CAMSHIFT is based
on colors, thus it requires the availability of color histogram
for the objects being tracked the desired objects within the
video sequences. As mentioned earlier, the color model was
built in the HSV domain on the basis of the hue component.
The spatial mean value is computed after selecting the search
window size and its initial position. The search window is
then centered in that value. The centroid and first-order
moment for x and y are then calculated. The process is
repeated again and again till convergence is achieved.
As the hand is tracked a contour is mapped with the
corresponding hand which further extracts a corresponding
convex hull (area). The recognition has been done through
modeling of the hand by mapping it to the number of defect
formed in it. Afterwards system tracks the numerous defects
generated by the hand and maps it to a meaningful command.
The gesture recognition system tracks the gestures performed
by the hand and counts the number of defects formed which it
maps to the appropriate action (numerical value) assigned in
the gesture vocabulary.