Center bias enhanced visual saliency detection method

Abstract

Due to the advancement in multimedia technology, the images and videos play a major role in day today life. How the humans are looking into the image? The computational models of visual attention used in many of the computer vision tasks such as image segmentation, object recognition, image understanding, etc. The proposed method aims to construct visual saliency model with the help of the center bias and pre-attentive features. The bottom-up visual saliency model is introduced. Influence of Center bias is the key mechanism of the proposed method. Difference of Gaussian (DOG) filter and gaussian envelope function are used to identify the salient objects of an image. The experimental results of the proposed method compared with five state-of-art-methods. The benchmark database is used to obtain the performance metrics. Receiver Operating Characteristics (ROC), precision, recall and f-measure are found to analyze the performance of the proposed method.