Bio-inspired visual attention model based on cognitive approach for indoor object detection

A biologically inspired computational model is de veloped which includes both training and attention phase. The low-level features such as color, intensity, orientation and textural information are extracted from the objects in the training phase. These following features are represented by mean and standard deviation which are stored in the memory. The similar features are extracted in the attention phase from the input image and it is used to construct final top-down saliency map. The proposed system introduces combined operation of Gabor wavelet transform and local binary pattern operator to extract texture features which are incorporated into computational architecture for constructing final saliency map. The model improves the detection and localization accuracy of indoor objects and it is tested for different indoor objects. The Gabor based textural pattern attains lower hit number and better detection rate for different object sets.