We described the face recognition algorithm adaptive principal component
analysis (APCA) and rotated adaptive principal component analyses (RAPCA),
which are insensitive to illumination and expression variations. We then extend
our previous work to multi view face recognition by interpreting facial features and
synthesizing realistic frontal face images when given a single novel face image. The experimental results show
that after frontal pose synthesis, the recognition rate increases significantly,
especially for larger rotation angles.
Furthermore, we examined how an automated face recognition system can be implemented on embedded
systems. We also explored various design approaches. We currently have two
prototype systems for the real-time automated face recognition.
The first prototype was entirely implemented on an Analog Devices Black fin DSP
processor capable of verifying a face from a database of 16 faces under a second. This was done as a
replacement for PIN identification on a NOKIA mobile phone. The second
prototype was developed using a hardware-software approach on a NIOS II processor with
extended instructions. The NIOS II processor was configured on an Alter FPGA.
There are several new directions that study in the area of
collection development for design and applied arts programs may take. One is
the growing importance of evidence-based design in several disciplines. It may
mean that design and applied arts programs will increase their scholarly output
in addition to their creative output. If nothing else, it will mean an
increased demand for library resources—and the need for help finding and using
those resources—as researchers look for evidence to support their design
decisions. Materials about sustainable and green design will continue to be of
great interest to fashion and interior designers. In addition, technology for
working with and presenting images will no doubt continue to improve, so face recognition software
No comments:
Post a Comment