Abstract:
Image processing technique
is used to uniquely identify the objects from an
image. In that way our research is based upon integrating an efficient image
processing technique to uniquely identify face from digital image sources.
Cross correlation, template matching and neural network are integrated to
form an efficient technique to identify faces. Firstly, the location of the face
will be roughly sketched using cross correlation from two templates one
with brighter background and another one with darker background. Then
Searching algo rithm in the neural network will recheck and confirm already
roughly plotted location. Not only it checks for the peak location, but also it
checks all the area around already plotted location. If the output of the
network is returned 0.5 for a search loca tion, all the pixel locations will be
marked around that pixel to be checked in the next search iteration. The
proposed system will be able to identify different types of digital images.
This system can be used for some automated statistical analytical pur poses.
This system will speed up the process of counting than the manual system
of counting. The proposed system will be developed using math lab and
open CV.