A new method for the extraction of Region-Of-Interest based on visual attention

摘要

Region of interest (ROI) plays an important role in image analysis, yet traditional ROI extraction result may be different from human's subject feeling. In this paper, a new approach based on visual attention is proposed considering that primates have remarkable abilities to interpret complex scenes real-time. This approach uses visual attention model to locate salient points. As a result, the most salient region is selected and marked as ROI. The proposed work is based on bottom-up influences of attention and, therefore, purely unsupervised. Furthermore, experiment results of the ROI extraction mentioned in the paper are given and show perfect performance when comparing with other traditional approaches.