Abstract:

 In order to evaluate target camouflage effectiveness objectively, precisely and conveniently, we establish a model based on the human visual attention mechanism. A conspicuity map is obtained based on intensity of the input image, orientation characteristics and chromatic aberration. Shift of the focus of attention (FOA) is selected and measured with the mechanisms of winner-take-all (WTA) and inhibition of return. The relation between the FOA shift and target detection probability is obtained using a BP neural network. Target camouflage effectiveness can be measured precisely. Experiments are performed on 44 images of southern forest for training the BP neural network and validating the model. The results show that this
model has advantages in precision, determinability and operability.

Key words: visual attention mechanism, camouflage effectiveness, evaluation, focus of attention