04.01.2015 Views

Fusion of Visual and Thermal Face Recognition Techniques: A ...

Fusion of Visual and Thermal Face Recognition Techniques: A ...

Fusion of Visual and Thermal Face Recognition Techniques: A ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

component is connected based on a chain coding scheme; finally, ellipses are fitted based<br />

on each connected component. Although face regions are not well representative, eye<br />

blobs are well fitted. This ellipse fitting method may be helpful to reduce the search time<br />

for detecting the eyes even in visual images.<br />

Figure 12: Ellipse fitting demo<br />

Detecting <strong>and</strong> tracking <strong>of</strong> face-like objects in cluttered scenes is an important<br />

preprocessing stage <strong>of</strong> overall automatic face recognition. The goal <strong>of</strong> face detection is to<br />

segment out face-like objects from cluttered scenes. Recent survey papers on face<br />

detection techniques in visual images can be found in [55][56]. The detection <strong>of</strong> faces in<br />

thermal images can be achieved by applying thresholding in which most skin areas are<br />

covered.<br />

Figure 13 shows a basic flow diagram <strong>of</strong> the face detection algorithm. Original<br />

images are binarized with a statistically driven threshold <strong>and</strong> processed by morphological<br />

(opening & closing) to reduce small blobs inside <strong>of</strong> the faces. Then the results are linked<br />

using a Freeman chain-coding algorithm. A small number <strong>of</strong> connected components is<br />

29

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!