Extended Non-Local Feature for Visual Saliency Detection in Low Contrast Images

Abstract

Saliency detection model can substantially facilitate a wide range of applications. Conventional saliency detection models primarily rely on high level features from deep learning and hand-crafted low-level image features. However, they may face great challenges in nighttime scenario, due to the lack of well-defined feature to represent saliency information in low contrast images. This paper proposes a saliency detection model for nighttime scene. This model is capable of extracting non-local feature that is jointly learned with local features under a unified deep learning framework. The key idea of the proposed model is to hierarchically introduce non-local module with local contrast processing blocks, aiming to provide robust representation of saliency information towards low contrast images with low signal-to-noise ratio property. Besides, both nighttime and daytime images are utilized in training to provide complementary information. Extensive experiments have been conducted on five challenging datasets and our nighttime image dataset to evaluate the performance of the proposed model.