Purchase this article with an account.
Timothy Dixon, Eduardo Canga, Stavri Nikolov, Tom Troscianko, Jan Noyes, Dave Bull, Nishan Canagarajah; Solider direction and soldier location: Image fusion and compression in two scene perception tasks. Journal of Vision 2007;7(9):201. doi: https://doi.org/10.1167/7.9.201.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The recent interest in image fusion - combining images of differing modalities such as visible light and infrared radiation - has created a growing body of research that indicates when fusion methods might be of more or less benefit, with the use of multiple assessment methods advised (e.g. Essock et al., 2004). The current work presents two such experiments, comparing different fusion methods (contrast pyramid [CP]; shift-invariant discrete wavelet transform [SIDWT]; and dual-tree complex wavelet transform [DT-CWT]) and JPEG2000 compression (no compression; low compression, .32bpp; high compression, .08bpp), as well as target location. In Experiment 1, participants indicated whether a soldier was facing left or right in six locations across a scene. In Experiment 2, they indicated whether a soldier was to the left or right of a clearing of trees in 12 locations across a scene. The results for Experiment 1 indicated no difference in fusion methods, but significant effects of compression and target location, with these two interacting. Reaction times (RTs) were slower with increased compression, and increased eccentricity of target, although RTs dropped for furthest targets, revealing an edge effect. Experiment 2 revealed main effects of fusion, compression and location, and an interaction between compression and fusion. RTs were faster for the SIDWT and DT-CWT than for CP, and slower between clean and high compression, with CP becoming much slower than the other fusion methods with increased compression. In addition, RTs slowed as target location approached the centre of the clearing, indicating an increase in uncertainty in participant response. Results indicate use of differing fusion schemes can benefit global target location but not more local target identification. However, compression has greater effect on identification rather than location. Future directions are presented, using an eye-tracking paradigm to further explore how fusion methods might be differentiated.
This PDF is available to Subscribers Only