Region-Based Top-Down Visual Attention Through Fine Grain Color Map

Z. Aziz and B. Mertsching

This paper is related to the top-down pathway of task driven attention using the behavior of visual search. Recent experiments on psychophyisics propose the fine-grain nature of top-down pathway. We present an attention model based upon this concept and investigate its ability to perform attention-oriented visual search in complex as well as structured visual input. The component related to the top-down color saliency map with a memory-based inhibition mechanism is explained here. Results of the model show success of the technique in terms of functionality, efficiency, and robustness.