OBJECT DETECTION IN NATURAL BACKGROUNDS PREDICTED BY DISCRIMINATION PERFORMANCE AND MODELS

Many models of visual performance predict image discriminability, the visibility of the difference between a pair of images, We compared the ability of three image discrimination models to predict the detectabi lity of objects embedded in natural backgrounds, The three models were : a multiple channel Cortex transform model with within-channel maskin g; a single channel contrast sensitivity filter model; and a digital i mage difference metric, Each model used a Minkowski distance metric (g eneralized vector magnitude) to summate absolute differences between t he background and object plus background images, For each model, this summation was implemented with three different exponents: 2, 4 and inf inity. In addition, each combination of model and summation exponent w as implemented with and without a simple contrast gain factor, The mod el outputs were compared to measures of object detectability obtained from 19 observers, Among the models without the contrast gain factor, the multiple channel model with a summation exponent of 4 performed be st, predicting the pattern of observer d's with an RMS error of 2.3 dB , The contrast gain factor improved the predictions of all three model s for all three exponents, With the factor, the best exponent was 4 fo r all three models, and their prediction errors were near 1 dB. These results demonstrate that image discrimination models can predict the r elative detectability of objects in natural scenes. Published by Elsev ier Science Ltd.