Recently, Zhou et al. designed a two‐stream faster Region‐Convolutional neural networks (R‐CNN) model RGB‐N for color image splicing localization in CVPR2018. However, the RGB‐N locates spliced regions only at block‐level and ignores the entirety and inherent correlation of three channels. Therefore, an improved quaternion two‐stream R‐CNN model is proposed to solve these drawbacks: a mask branch combining fully convolutional network and condition random field is added for locating spliced regions at pixel‐level; quaternion representation of color images is used to process color spliced images in a holistical way. In addition, feature pyramid network based on quaternion residual network is considered to extract multi‐scale features for color spliced images; attention region proposal network is combined with attention mechanism and is designed to pay more attention to the spliced regions; a high‐pass filter designed for image splicing detection specifically is adopted to replace steganalysis rich model filter in the RGB‐N to obtain noise input for the noise stream. Experimental results on a new synthetic dataset and three standard forgery datasets demonstrate that the proposed method is superior to the existing methods in the abilities of localization, generalization, and robustness.
Structured-sparsity regularization is popular for sparse learning because of its flexibility of encoding the feature structures. This paper considers a generalized version of structured-sparsity regularization (especially for $l_1/l_{\infty}$ norm) with arbitrary group overlap. Due to the group overlap, it is time-consuming to solve the associated proximal operator. Although Mairal~\shortcite{mairal2010network} have proposed a network-flow algorithm to solve the proximal operator, it is still time-consuming especially in the high-dimensional setting. To address this challenge, in this paper, we have developed a more efficient solution for $l_1/l_{\infty}$ group lasso with arbitrary group overlap using an Inexact Proximal-Gradient method. In each iteration, our algorithm only requires to calculate an inexact solution to the proximal sub-problem, which can be done efficiently. On the theoretic side, the proposed algorithm enjoys the same global convergence rate as the exact proximal methods. Experiments demonstrate that our algorithm is much more efficient than network-flow algorithm, while retaining the similar generalization performance.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.