2006 CPCUG Science Fair Awards

On May 8th, 2006, General Meeting CPCUG recognized two young computer scientists. Since 1996 the Capital PC User Group has honored deserving young computer scientists for their outstanding science fair projects. CPCUG President Dennis Courtney presented the awards including present plaques, $50 savings bonds, and one-year CPCUG memberships for special achievement in exploring computer technology.

Loudoun County Science and Engineering Fair Awardees

Name and Project Title	Abstract
Sean Guarino I See You: The Creation of a Computer-based Facial Identification Program	Recognition programs using biometrics are increasingly being developed in order to provide greater security measures to business and law enforcement.A facial recognition program was created to explore whether or not it was possible to identify faces using a web camera. The use of a web cam would be more economical than cameras typically used today. Four images of twenty subjects were made using a web cam and cropped to 80 x 80 pixels to show only the face. The image data was loaded into the identification program and one image selected from the pool of twenty. Three trials were conducted for each individual, who was correctly identified with at least 90% accuracy. Further research would entail increasing the accuracy of the program, which may also have application in animal tracking and endangered species identification.
Jessica A Shipman The Relationship Between Image Compression Type and Image Quality	A common problem that many digital photographers face is how to store their images. The many types of image compression, and the tradeoff between image quality and file size, can puzzle the user. This experiment used three pictures of different subjects (a human, an artificial structure and a natural scene), a text and a graphic image. These were processed using nine compression methods and grayscale. The variance, entropy and mean of the compressed images were evaluated using Mathematica. The hypothesis was that JPEG would be the best compression for photographs and PNG would be best for text and graphic images. The results show PNG is best for all types of images when quality is most important. The next best methods were GIF for text and grayscale images, and JPEG for photographs and graphics. Further research is needed to investigate the relationship between compression method, image quality and file size, to find the optimal method considering both quality and storage space.

Name and Project Title

Abstract

Sean Guarino

I See You: The Creation of a Computer-based Facial Identification Program

Recognition programs using biometrics are increasingly being developed in order to provide greater security measures to business and law enforcement.A facial recognition program was created to explore whether or not it was possible to identify faces using a web camera. The use of a web cam would be more economical than cameras typically used today. Four images of twenty subjects were made using a web cam and cropped to 80 x 80 pixels to show only the face. The image data was loaded into the identification program and one image selected from the pool of twenty. Three trials were conducted for each individual, who was correctly identified with at least 90% accuracy. Further research would entail increasing the accuracy of the program, which may also have application in animal tracking and endangered species identification.

Jessica A Shipman

The Relationship Between Image Compression Type and Image Quality

A common problem that many digital photographers face is how to store their images. The many types of image compression, and the tradeoff between image quality and file size, can puzzle the user. This experiment used three pictures of different subjects (a human, an artificial structure and a natural scene), a text and a graphic image. These were processed using nine compression methods and grayscale. The variance, entropy and mean of the compressed images were evaluated using Mathematica. The hypothesis was that JPEG would be the best compression for photographs and PNG would be best for text and graphic images. The results show PNG is best for all types of images when quality is most important. The next best methods were GIF for text and grayscale images, and JPEG for photographs and graphics. Further research is needed to investigate the relationship between compression method, image quality and file size, to find the optimal method considering both quality and storage space.

Our thanks go to CPCUG member George Herbert who judged in Loudoun County. It's challenging to understanding and assess the projects and rewarding to see these young people doing such good work. This program is an excellent way to bring new and energetic members into CPCUG. Moreover, the awards provides CPCUG with positive visibility with the scientifically literate families of the Washington DC area. If you would like to see your jurisdiction included in next year's effort, please contact Walt Houser CPCUG Coordinator at.

CPCUG Science Fair Home Page