Random walks on graphs to model saliency in images

Abstract

We formulate the problem of salient region detection in images as Markov random walks performed on images represented as graphs. While the global properties of the image are extracted from the random walk on a complete graph, the local properties are extracted from a k-regular graph. The most salient node is selected as the one which is globally most isolated but falls on a compact object. The equilibrium hitting times of the ergodic Markov chain holds the key for identifying the most salient node. The background nodes which are farthest from the most salient node are also identified based on the hitting times calculated from the random walk. Finally, a seeded salient region identification mechanism is developed to identify the salient parts of the image. The robustness of the proposed algorithm is objectively demonstrated with experiments carried out on a large image database annotated with 'ground-truth' salient regions.

Salient Object Detection: A Benchmark

“…In Hou and Zhang [5], the gist of the scene is represented with the average Fourier envelope and the differential spectral components are used to extract salient regions. This is replaced by the phase spectrum of the Fourier transform in [34] because it is more effective and computationally efficient. Some researchers including Bruce and Tsotsos [4] and Zhang et al [10] attempted to define visual saliency based on information theory.…”

Saliency propagation from simple to difficult

“…However, such propagations may incur errors if the unlabeled adjacent superpixels are inhomogeneous or very dissimilar to the labeled ones. For example, [7] and [10] formulate the saliency propagation process as random walks on the graph. [21] and [27] conduct the propagations by employing personalized PageRank [29] and manifold based diffusion [29], respectively.…”

Visual saliency-based active learning for prostate magnetic resonance imaging segmentation

“…Visual saliency models selecting the most salient region in an image are based on, among others, local contrast, 24 entropy, 25 and graph-based random walks. 26,27 High-contrast regions have maximum information and hence higher entropy. 25 Regions with high classification uncertainty also have high entropy, indicating a correspondence between information content of salient regions and classification uncertainty.…”