Abstract:Saliency detection is an important and challenging research topic due to the variety and complex of the background and saliency regions. In this paper, we present a novel unsupervised saliency detection approach by exploiting a learning-based ranking framework. First, the local linear regression model is adopted to simulate the local manifold structure of every image element, which is approximately linear. Using the background queries from the boundary prior, we construct a unified objective function to globally minimize all the errors of the local models for the whole image element points. The Laplacian matrix is learned via optimizing the unified objective function. Low-level image features as well as high-level semantic information extracted from deep neural networks are used for the Laplacian matrix learning. Based on the learnt Laplacian matrix, the saliency of the image element is measured as the relevance ranking to the background queries. The foreground queries are obtained from the background-based saliency and the relevance ranking to the foreground queries is calculated in the same way as the background-based saliency. Second, we calculate an enhanced similarity matrix by fusing two different-level deep feature metrics through cross diffusion. A propagation algorithm uses this enhanced similarity matrix to better exploit the intrinsic relevance of similar regions and improve the saliency ranking results effectively. Results on four benchmark datasets with pixel-wise accurate labelling demonstrate that the proposed unsupervised method shows better performance compared with the newest state-of-the-art methods and is competitive with deep learning-based methods.

Nonlocal Variational Model for Saliency Detection

Saliency Detection Via Local Single Gaussian Model

Saliency Detection via Nonlocal Minimization

Saliency Detection via Nonlocal $$L_{0}$$ Minimization

A Nonlocal $$L_{0}$$ L 0 Model with Regression Predictor for Saliency Detection and Extension

Saliency Detection by Non-Linear Intensity Mapping in Images

Visual Saliency Detection Based on Topographic Independent Component Analysis

Nonparametric saliency detection using kernel density estimation

Graph Laplacian Based Visual Saliency Detection

A Saliency Detection Model Using Low-Level Features Based on Wavelet Transform

A Hybrid of Local and Global Saliencies for Detecting Image Salient Region and Appearance

Object-Based Visual Saliency Via Laplacian Regularized Kernel Regression

Saliency detection using parallel non-linear integration of color and gradient using covariances

Visual Saliency Detection Using Local Patches Contrast

Local Regression Ranking for Saliency Detection

Nonlocal Feature Transform with Iterative Refinement for Video Saliency Detection

A Saliency Detection Model Combined Local And Global Features

A Novel Non-Local Geodesic Active Contour Model

A Computational Model for Saliency Maps by Using Local Entropy

A biologically inspired computational model for image saliency detection.

Improvement of Salient-Region Detection Using an Integrated Bottom-Up Model