Showing papers in "Journal of Visual Communication and Image Representation in 2018"

TL;DR: Experiments show that the SFCN and MFCN outperform existing splicing localization algorithms, and that the M FCN can achieve finer localization than the S FCN.

TL;DR: A dataset of color images corrupted by natural noise due to low-light conditions is introduced, together with spatially and intensity-aligned low noise images of the same scenes, and a method for estimating the true noise level in the authors' images, since even the low noise image contain small amounts of noise.

TL;DR: In this paper, a Region Ensemble Network (REN) is proposed to exploit good practice and promote the performance for hand pose estimation, which first partitions the last convolutional outputs of ConvNet into several grid regions.

TL;DR: Experimental results show that the ATTention U-Net (ATT-UNet) achieves consistent improvement in both visible wavelength and near-infrared iris images with challenging scenery, and surpass other representative iris segmentation approaches.

TL;DR: The proposed forgery detection technique can be applied to detect the tampered areas and the benefits can be obtained in image forensic applications.

TL;DR: Experimental results, based on the publicly available OUC-VISION underwater image database, show that the proposed method can produce reliable and promising results, compared to other state-of-the-art saliency-detection models.

TL;DR: In this paper, a data-driven Saak transform with augmented kernels is proposed, which consists of three steps: (1) building the optimal linear subspace approximation with orthonormal bases using the second-order statistics of input vectors, (2) augmenting each transform kernel with its negative, and (3) applying the rectified linear unit (ReLU) to the transform output.

TL;DR: This paper proposes a novel shape decomposition-based segmentation technique to decompose the compound characters into prominent shape components, which reduces the classification complexity in terms of less number of classes to recognize, and at the same time improves the recognition accuracy.

TL;DR: The experimental results demonstrate that the proposed scheme is imperceptible and robust against a variety of intentional or unintentional attacks.

TL;DR: Subjective as well as objective evaluations demonstrate that the fusion quality in terms of edge strength, standard deviation, feature mutual information, fusion factor, feature similarity and structural similarity has significantly improved in the proposed algorithm as compared to other state-of-art multimodal medical image fusion algorithms.

TL;DR: A new median filtering detection method based on CNN is proposed and achieves significant improved detection performance and performs well for highly compressed image of size as small as 16 × 16.

TL;DR: A novel end-to-end learnable LBP network for face spoofing detection that substantially outperforms the state-of-the-art methods and can significantly reduce the number of network parameters.

TL;DR: A novel computational model for saliency detection is proposed by integrating the holistic center-directional map with the principal local color contrast (PLCC) map, which is sufficient to highlight and separate salient objects from complex background while dramatically reduce the computational cost.

TL;DR: Different studies of the visual features for FER are brought together by evaluating their performances under the same experimental setup, and critically reviewing various classifiers making use of the local descriptors.

TL;DR: Cutting-edge data-driven concepts and deep convolutional neural networks are leveraged to harness enough characterization aspects from a wide range of images and point out the presence of child pornography content in an image.

TL;DR: Experiments on four public benchmark databases indicate that the proposed face spoofing detection scheme can effectively resist photo and video spoofing attacks in face recognition.

TL;DR: Experimental results demonstrate that the proposed steganography algorithm achieves higher embedding capacity with better imperceptibility compared to the published steganographic methods.

TL;DR: A scalable pipeline for Free-Viewpoint Video content creation that incorporates bio-mechanical constraints through 3D skeletal information as well as efficient camera pose estimation algorithms and introduces multi-source shape-from-silhouette combined with fusion of different geometry data as crucial components for accurate reconstruction in sparse camera settings.

TL;DR: A center surround filter is employed to improve speed and memory requirements of the transmission estimation in image dehazing, and is compared with that of other state of the art methods using a subjective quality assessment method and a number of objective quality assessment methods.

TL;DR: An approach integrating visual saliency model with BOW is proposed for semantic image retrieval and the results evaluated in terms of mean Average Precision show that this proposal outperforms the referred state-of-the-art approaches.

TL;DR: The CNN-GRNN model replace Back propagation neural network inside CNN with GRNN to improve generalization and robustness of CNN and it is shown that the model is superior to Gray Level Co-occurrency, HU invariant moments, CNN and CNN_SVM on small sample dataset.

TL;DR: Evaluated methods highlight the effectiveness of the proposed method in its ability to produce output maps that are clear and readable, and which can achieve successful detections on cases where other algorithms fail.

TL;DR: A novel deep learning based approach, in which the Convolution Neural Network and Recurrent Neural Network are used to solve the specific problem of detecting copied segments in videos to achieve significant performance improvements compared to the state of the art.

TL;DR: An edge-preserving smoothing pyramid is introduced for the multi-scale exposure fusion and the experimental results prove that the proposed algorithms produce better fused images than the state-of-the-art algorithms both qualitatively and quantitatively.

TL;DR: Significant feature extraction approaches and clustering methods applied on the image data from nine important applicative areas are reviewed and characteristics of images, suitable clustering approaches for each domain, challenges and future research directions for image clustering are discussed.

TL;DR: This paper proposes an efficient and straightforward approach, video you only look once (VideoYOLO), to capture the overall temporal dynamics from an entire video in a single process for action recognition.

TL;DR: This paper proposed a face-mask recognition method for fraud prevention based on Gaussian Mixture Model, which is simple to be calculated and has a higher accuracy, and enhanced the robustness of the algorithm about mask recognition.

TL;DR: It was found that percentile values computed from sorted, level-shifted, high-frequency wavelet coefficients can serve as reliable image sharpness/blurriness estimators and higher dynamic range of contrast maps significantly improves model performance.

TL;DR: This study sought to develop a deep learning model to automatically detect the diagnostic images of Confocal laser endomicroscopy and explored the effect of training regimes and ensemble modeling and localized histological features from diagnostic CLE images.

TL;DR: Experimental results show that the proposed method for JPEG images using adaptive embedding has a better performance than state-of-the-art works and the file size can be well preserved after data hiding.