Abstract
This study researches the coding model adaptive for information processing of the bottom-up attention mechanism. We constructed a coding model satisfying the neurobiological constraints of the primary visual cortex. By quantitatively changing the coding constraints, we carried out experiments on images used in cognitive psychology and natural image sets to compare the effects on the saliency detection performance. The experimental results statistically demonstrated that the encoding of invariant features and representation of overcomplete bases is advantageous to the bottom-up attention mechanism.
Highlights
1.1 Computational models for bottom-up attentionBottom-up attention models extract multi-dimensional features from an image and combine these features into a saliency map where the most salient object will be perceived.In the feature extraction stage, computational models motivated by imitation of the primary visual cortex often use Gabor filters to extract orientation information at different scales
Does invariant representation in the primary visual cortex affect bottom-up attention? Second, what is the effect of overcomplete representation on saliency detection?
By quantitatively changing the coding constraints, we conducted experiments on images used in cognitive psychology and natural image sets to compare the effects on the saliency detection performance caused by the different coding constraints
Summary
1.1 Computational models for bottom-up attentionBottom-up attention models extract multi-dimensional features from an image and combine these features into a saliency map where the most salient object will be perceived.In the feature extraction stage, computational models motivated by imitation of the primary visual cortex often use Gabor filters to extract orientation information at different scales. Bottom-up attention models extract multi-dimensional features from an image and combine these features into a saliency map where the most salient object will be perceived. In the feature extraction stage, computational models motivated by imitation of the primary visual cortex often use Gabor filters to extract orientation information at different scales. Properties of Gabor filters resemble simple cells’ receptive fields and can provide input to the bottomup saliency map. Similar methods use Gaussian pyramids [13], Fourier transformation, or wavelets decomposition [10] to extract features similar to the responses of cells. One of representative models proposed by Itti et al [3] adopted Gaussian pyramids to extract color, intensity, and orientation features at different levels. Grigorescu’s model [14] simulated complex cells and nonclassical receptive field inhibition to detect salient contours
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
