Face Recognition and Eye Blinking System
Spring 2004, Carnegie Mellon University

The goal of this project was to create a face recognition system with eye-blinking detection capabilities. It would be applied to a security system, where instead of using a key, magnetic card, or any other conventional method, a person would gain access by blinking his or her eyes in front of a camera. By blinking ones eyes in a specific sequence, our system would detect the sequence, convert it into a binary code, and compare it with entry codes already pre-programmed in our database, provided of course that the person has been previously identified by our face recognition system. Using Altera's NIOS development kit and a PIC processor, the system had to be implemented mostly on the FPGA that would make use of multiple clock domains. Our system had to be extremely concise and restricted since it had to have minimum memory capabilities and processor power.

Proceed to details

Hand Gesture Recognition
Spring 2004, Carnegie Mellon University

Sign language is one form of communication for the hearing and speech impaired. Similar to spoken language, there is no universal sign language. Sign language is itself a separate language with its own grammar and rules. Some signs are expressed as static gestures while others incorporate some dynamic hand movements. For static gestures, the prominent sign is captured within a specific time frame. For dynamic gestures, a sequence of finger and hand positions needs to be identified and analyzed in order to be recognized.The focus of this project is on gestures with a single hand. We strive to detect a hand signing the sign language representation of the numbers from 0 to 9. Further on into the project, I included a video stream with different signs, and the Matlab code had to detect the numbers on the fly. To see a video demonstration of the algorithm, click on the image.

Proceed to details

Image Processing Console
Summer 2003, Carnegie Mellon University

I'm staying here at Carnegie Mellon for part of the summer 2003, working with professor Takeo Kanade, as a research assistant. The research I'm working on right now is to create an image processing console, to analyze pictures taken from a surveillance camera, detect any human figure, apply noise reduction algorithms to it, and archive the almost 'perfect' human silhouette into a database. This database can be used in the future for real time matching of human figures from pictures taken from surveillance cameras. Possible improvements that can be added to it, would be zooming in on specific segments of the body and archive the shape, emotions, or charateristic details from every individual captured by the camera.