Effects Of Image Parameters On Erps And Verification Of Face Familiarity

Z Shen,W Huan,Pf Wu,Jf Li
1998-01-01
Abstract:Even though some ERP components related to face recognition have been reported since the 1980s, we do not know what component of face, edge, contour, structural parts or contexture is the most important to face familiarity. The paper presented an ERP study by image processing techniques.ERPs in 26 college students were elicited by 10 categories of face photos displayed on a screen through a projector. Each group of five of familiar and strange face photos included: natural, lowpass and high-pass filtering, and those processed by Laplacian transform (LT) and Robert transform (RT), respectively. All of the processed photos were made by an image processor (Sun 316, DEC Co.).Unprocessed familiar and strange photos elicited significantly different P300; high-pass photos made the verification worse and the P300 difference disappear. LP photos did not hamper familiar judgment, although they delayed the latency of the N200 component. RT photos decreased the correct ratio of familiarity judgment and increased reaction time, while it delayed the latency of N200.The results suggested that the amplitude of P300 at Pt was sensitive to the familiarity of natural photos, but not to the processed photos and that the contexture was more important than edge or contour; the nonlinear feature in contours was more relevant than the linear feature for verifying familiarity of the face.
What problem does this paper attempt to address?