Saliency Detection Based on Human Perception of Visual Information

This paper proposes an algorithm for accurate detection of salient areas from a given scene. We used a complex model for the human visual system, in order to simulate the visual perception mechanisms. Human visual system modelling requires accurate knowledge about the entire visual pathways. This work focuses on the following features of the human vision: the color perception mechanism, the perceptual decomposition of visual information in multiple processing channels, contrast sensitivity, pattern masking and detection/pooling mechanism present in the primary visual cortex. Pattern masking is considered within a complex approach, combining data from distinct dimensions. The results are also shown to correlate with the subjective results obtained for several test sequences.