A DCT based in-focus visual saliency detection algorithm.

Abstract

A novel low-complexity visual saliency detection algorithm for detecting visually salient regions in images based on camera focus is presented. This fast in-focus region detection algorithm detects salient-frequencies present in in-focus areas using the characteristics of Discrete Cosine Transform coefficients. The performance of this algorithm is validated against five state-of-the-art saliency detection algorithms. The results show that this algorithm consistently outperforms other saliency detection algorithms in terms of complexity and prediction accuracy.