|
||||||
|
||||||
| Undergraduate Honours Projects | ||||||
|
Carleton University - School of Computer Science Undergraduate Honours Project Fall 2009 Facial Recognition on Mobile Platforms Terry Worona 
ABSTRACT The introduction of smart phones configured with cameras, GPS systems and internet browsers have ushered in a new era of information technology. The report details the creation of Look, an iPhone application which addresses the lack of integration between computer vision and mobile platforms. Look is designed to locate a face within an image captured by the iPhone camera and to retrieve a match from stored images in a remote database. The closest match is returned and displayed together with the original source image. The model database is created in an off-line process using an eigenspace algorithm which compresses images by storing them as eigenvector representations, while the actual face images are stored on a Flickr web server. The model database is compiled as an XML file and downloaded by Look at startup. As a match is located, the specific face image is returned to the iPhone via a Flickr JSON request. The report determines that it is possible to cross-compile OpenCV libraries for mobile ARM architectures. Contributions are noted in the conclusion, providing specific instructions for porting OpenCV to the iPhone, representing an image database as an eigenspace XML and querying REST services such as Flickr. |
||||||
|
