Skip to main content


Face recognition using a visual search paradigm

Trainee Achievements

Face recognition using a visual search paradigm

The ability to recognize faces is fundamental to both human perception and computer vision. Computer algorithms have been developed whose performance approaches that of humans, but significant questions remain about how robust these procedures will be in the presence of image degradations that occur under real-life conditions. Trainees Denisova, Haladjian, Kibbe and Mansley developed a paradigm to study face recognition using a visual search paradigm. Faces were created and features were analyzed using open source face-recognition algorithms available on OpenCV. They found that the reaction time of observers to find a target face among distractors was degraded by image distortion (Gaussian blur), but effects of noise could be ameliorated considerably by imposed motion. The beneficial effects of motion found in human perceivers could inform the development of automated face-detection software.