Saliency Detection: A Spectral Residual Approach

TLDR

The human visual system detects visual saliency quickly and reliably, yet computational modeling of this ability remains difficult. This paper proposes a simple method for visual saliency detection. The model, which does not rely on features, categories, or prior knowledge, analyzes the log‑spectrum of an input image to extract its spectral residual and then constructs a fast saliency map in the spatial domain, and it is tested on both natural and artificial images. Results show that the method provides fast and robust saliency detection.

Abstract

The ability of human visual system to detect visual saliency is extraordinarily fast and reliable. However, computational modeling of this basic intelligent behavior still remains a challenge. This paper presents a simple method for the visual saliency detection. Our model is independent of features, categories, or other forms of prior knowledge of the objects. By analyzing the log-spectrum of an input image, we extract the spectral residual of an image in spectral domain, and propose a fast method to construct the corresponding saliency map in spatial domain. We test this model on both natural pictures and artificial images such as psychological patterns. The result indicate fast and robust saliency detection of our method.