Abstract
Engineering, University of Manchester, UK; e-mail: [email protected]) There is a tendency for observers to look initially at the centre of an image [eg Tatler, 2007 Journal of Vision 7(14):4, 1 ^ 17]. The effect has been attributed variously to optimal information processing strategies and artistic bias in test-image selection. It is not clear, however, whether target-detection performance is similarly biased. To address this question, a target-detection task was undertaken with coloured images containing a target at one of 130 possible locations and matched in luminance to its local surround. Each image, subtending approx. 17613 deg at 1 m, was viewed for 1 s on a computer screen. In all, 20 natural rural and urban scenes were tested. The target was a grey sphere (Munsell N7), subtending approx. 0.2 deg. Detection performance was quantified by the index d 0 and plotted against target location. A central bias was apparent in performance averaged over scenes, but it disappeared with most individual scenes. For many scenes, d 0 correlated positively with local luminance. As a control, when sections of the image were permuted, the position of peak d 0 migrated with the permutation. For the natural scenes tested here, observers seem better at detecting targets where the image is brighter. [Supported by EPSRC grant no. EP/F023669/1.]
Original language | English |
---|---|
Pages | 180-180 |
Number of pages | 1 |
Publication status | Published - Aug 2009 |
Event | ECVP 2009 - Regensberg, Germany Duration: 24 Aug 2009 → 28 Aug 2009 |
Conference
Conference | ECVP 2009 |
---|---|
City | Regensberg, Germany |
Period | 24/08/09 → 28/08/09 |