RIC
Regional Information Centre
for Scientific and Technological Cooperation with EU
 

FP7 Partner offer profile

Kursk State Technical University - HUMAN EYES RECOGNITION FOR DIGITAL IMAGE AUTOMATIC FRAMING

Go back
Date of filling in 20.02.2008
Organization name Kursk State Technical University
Organization adress 50 let Oktyabrya St., 94, Kursk-40, 305040
Department/Unit Chair of Computer Engineering
Contact person Titov Vitaliy, DSc, Professor, Head of Chair
Phone +7-4712-587105
Fax +7-4712-587112
E-Mail titov@vt.kstu.kursk.ru
tmi@pub.sovtest.ru
Web-site http://www.kstu.kursk.ru
Organization type
  • Research
  • Education
Special Programme
Theme
Information and Communication Technologies (ICT)
Call identifier
Topic number HUMAN EYES RECOGNITION FOR DIGITAL IMAGE AUTOMATIC FRAMING
Call Deadline
Short description of the organization (max 12 lines): -
Expertise offered Nowadays digital cameras already are not the luxury that is accessible to a small group of professional photographers. A wide range of inexpensive digital cameras, manufactured by the leading world vendors (Sony, Kodak, Olympus, Nikon, Samsung and etc), is available on the market. Optics quality, CCD matrix resolution and physical dimensions, cost of cameras of different vendors are close, so the choice of a concrete model depends on its functionality that include image processing capabilities. Our goal is to develop a method for automatic portrait digital images framing with human face centering in a frame plane that will be embedded to digital cameras. The following results are obtained: The developed method is:
  1. Capable to detect human eyes on a digital image with resolution greater then 320x240 pixels independently from a face position in a frame plane (is invariant to a shift and rotation of a face in a frame plane).
  2. Is invariant to a change of distance between a camera and human and, as a result, to a change of face scale on an image. The method provides a reliable recognition when eyes linear sizes are 40 pixels or greater.
  3. Is designed for eyes recognition on images with good contrast.
Problems to be solved and results: The subsequent investigations will be aimed to:
  • Provide eyes recognition of a group of people.
  • Lower the requirements to minimal image sizes and minimal eyes linear sizes.
  • Increase method efficiency on images with bad contrast that are obtained without artificial illumination.
  • Provide animals eyes recognition.
The developed eyes recognition method is designed for digital cameras and provides an automatic best frame selection from a sequence based on criterion of opened eyes and automatic portrait images framing. It is applicable in ophthalmology for automatic strabismus diagnostics and in biometric access control systems based on human face or iris pattern recognition.
Scientific keywords Machine Vision Image processing Eyes recognition Automatic framing Digital image
Publications on the topic (other references) -
Description of previous and present experience in International Cooperation -
Previous participation in EU’s Framework Programme projects