Fusion Visual Attention and Low-Level Features in Images for Region of Interest Extraction

For the point of Region of interest (ROI) extraction influenced by low-level features in images, this paper proposes two ROI extraction algorithms. The first one is based on eye movement data and use proposed “Marking Helm” method. The second one is an optimal weighted feature ROI extraction algorithm based on performance evaluation of feature ROI with different low-level features such as color, intensity, orientation and texture feature in images. By analyzing the similarity between the extracted ROIs by the proposed optimal weighted feature ROI extraction algorithm, visual attention models and eye movement data respectively, this paper confirms the validity of the proposed algorithm. Experimental results show that the proposed optimal weighted feature ROI extraction algorithm improves the similarity at least ten percent over the ROI extraction algorithm based on Itti and Stentiford visual attention models.