Search

Face hallucination is a super‐resolution technique specially designed to reconstruct high‐resolution faces from low‐resolution faces. Most state‐of‐the‐art algorithms leverage position‐patch prior knowledge of human faces to better super‐resolve face images. However, most of them assume the training face dataset is sufficiently large, well cropped or aligned. This paper, proposes a novel example‐based face hallucination method, based on cluster consistent dictionary learning with the assumption that human faces have similar facial structures. In this method, the paired face image patches are firstly labelled as face areas including eyes, nose, mouth and other parts, as well as non‐face areas without requiring the training face images cropped and aligned. Then, the training patches are clustered according their labels and textures. The cluster consistent dictionary is learned to represent the low‐resolution patches and the high‐resolution patches. Finally, the high‐resolution patches of the input low‐resolution face image can be efficiently generated by using the adjusted anchored neighbourhood regression. As utilizing the labelled facial parts prior knowledge, the proposed method represents more details in the reconstruction. Experimental results demonstrate that the authors' algorithm outperforms many state‐of‐the‐art techniques for face hallucination under different datasets. [ABSTRACT FROM AUTHOR]

Studying representation learning and generative modelling has been at the core of the 3D learning domain. By leveraging the generative adversarial networks and convolutional neural networks for point‐cloud representations, we propose a novel framework, which can directly generate 3D objects represented by point clouds. The novelties of the proposed method are threefold. First, the generative adversarial networks are applied to 3D object generation in the point‐cloud space, where the model learns object representation from point clouds independently. In this work, we propose a 3D spatial transformer network, and integrate it into a generation model, whose ability for extracting and reconstructing features for 3D objects can be improved. Second, a point‐wise approach is developed to reduce the computational complexity of the proposed network. Third, an evaluation system is proposed to measure the performance of our model by employing various categories and methods, and the error, considered as the difference between synthesized objects and raw objects are quantitatively compared, is less than 2.8%. Extensive experiments on benchmark dataset show that this method has a strong ability to generate 3D objects in the point‐cloud space, and the synthesized objects have slight differences with man‐made 3D objects. [ABSTRACT FROM AUTHOR]

Recently, with the development of various camera sensors and internet network, the volume of digital images is becoming big. Content-based image retrieval (CBIR), especially in network big data analysis, has attracted wide attention. CBIR systems normally search the most similar images to the given query example among a wide range of candidate images. However, human psychology suggests that users concern more about regions of their interest and merely want to retrieve images containing relevant regions, while ignoring irrelevant image areas (such as the texture regions or background). Previous CBIR system on user-interested image retrieval generally requires complicated segmentation of the region from the background. In this paper, we propose a novel hierarchical-local-feature extraction scheme for CBIR, whereas complex image segmentation is avoided. In our CBIR system, a perception-based directional patch extraction method and an improved salient patch detection algorithm are proposed for local features extraction. Then, color moments and Gabor texture features are employed to index the salient regions. Extensive experiments have been performed to evaluate the performance of the proposed scheme, and experimental results show that the developed CBIR system produces plausible retrieval results. [ABSTRACT FROM AUTHOR]

Highlights • A hierarchical framework is proposed to mimic humans for handling image tags. • A tree structure is constructed for image labels using a learning method. • Our model treats simple labels and complex labels separately. • Experiments results show promising performance on different datasets. Abstract Artificial Intelligence has attracted much of researchers' attention in recent years. A question we always ask is: "Can machines replace human beings to some extent?" This paper aims to explore the knowledge learning for an image-annotation framework, which is an easy task for humans but a tough task for machines. This paper's research is based on an assumption that machines have two systems of thinking, each of which handles the labels of images at different abstract levels. Based on this, a new hierarchical model for image annotation is introduced. We explore not only the relationships between the labels and the features used, but also the relationships between labels. More specifically, we divide labels into several hierarchies for efficient and accurate labeling, which are constructed using our Associative Memory Sharing method, proposed in this paper. [ABSTRACT FROM AUTHOR]

In video surveillance, the captured face images are usually of low resolution (LR). Thus, a framework based on singular value decomposition (SVD) for performing both face hallucination and recognition simultaneously is proposed in this paper. Conventionally, LR face recognition is carried out by super-resolving the LR input face first, and then performing face recognition to identify the input face. By considering face hallucination and recognition simultaneously, the accuracy of both the hallucination and the recognition can be improved. In this paper, singular values are first proved to be effective for representing face images, and the singular values of a face image at different resolutions have approximately a linear relation. In our algorithm, each face image is represented using SVD. For each LR input face, the corresponding LR and high-resolution (HR) face-image pairs can then be selected from the face gallery. Based on these selected LR–HR pairs, the mapping functions for interpolating the two matrices in the SVD representation for the reconstruction of HR face images can be learned more accurately. Therefore, the final estimation of the high-frequency details of the HR face images will become more reliable and effective. The experimental results demonstrate that our proposed framework can achieve promising results for both face hallucination and recognition. [ABSTRACT FROM AUTHOR]

Establishing correct correspondences between two faces with different viewpoints has played an important role in 3D face reconstruction and other computer-vision applications. Usually, face images are considered to lack sufficient distinctive features to establish a large number of correspondences on uncalibrated images. In this paper, we investigate pore-scale facial features, which are formed from pores, fine wrinkles, and hair. These features have many characteristics that make them suitable for matching facial images under different variations. Using both biological observation and computer-vision consideration, a new framework is devised for pore-scale facial-feature extraction and matching. The matching difficulty under various skin appearances of different subjects and imaging distortion is also analyzed. For further improving the matching performance and tackling distortions such as varying illuminations and unfocused blurring, a pore-to-pore correspondences dataset is established for training a more distinctive and compact descriptor. Experiments are conducted on a face database containing 105 subjects, and the results prove that the pore-scale features are highly distinctive; face images with a minimum resolution of 600×700 (0.4 mega) pixels contain sufficient details to perform a reliable matching in different poses. Generally, our algorithm can establish between 500 and 2000 correct correspondences on a pair of uncalibrated face images of the same person. Furthermore, the proposed methods can be applied to face recognition, 3D reconstruction, etc. [ABSTRACT FROM AUTHOR]

The one-sample-per-person problem has become an active research topic for face recognition in recent years because of its challenges and significance for real-world applications. However, achieving relatively higher recognition accuracy is still a difficult problem due to, usually, too few training samples being available and variations of illumination and expression. To alleviate the negative effects caused by these unfavorable factors, in this paper we propose a more accurate spectral feature image-based 2DLDA (two-dimensional linear discriminant analysis) ensemble algorithm for face recognition, with one sample image per person. In our algorithm, multi-resolution spectral feature images are constructed to represent the face images; this can greatly enlarge the training set. The proposed method is inspired by our finding that, among these spectral feature images, features extracted from some orientations and scales using 2DLDA are not sensitive to variations of illumination and expression. In order to maintain the positive characteristics of these filters and to make correct category assignments, the strategy of classifier committee learning (CCL) is designed to combine the results obtained from different spectral feature images. Using the above strategies, the negative effects caused by those unfavorable factors can be alleviated efficiently in face recognition. Experimental results on the standard databases demonstrate the feasibility and efficiency of the proposed method. [ABSTRACT FROM AUTHOR]

Abstract: A novel algorithm based on spatial and statistical information is proposed for the display of high-dynamic range (HDR) images. In our proposed algorithm, an image is first decomposed into a base layer and a detailed layer, which represent its smoothed and fine details, respectively. The problem of overall impression preservation is regarded as a global issue in our algorithm. Statistical-based histogram adjustment is employed to deal with the base layer. The reproduction of visual details is regarded as a local issue. The detailed layer obtained using a spatial filter is adaptively enhanced according to the mapping function used for the base layer. The main contributions of our algorithm are that: (1) an adaptive detail-enhancement method is proposed and (2) a gain map is defined to combine the local and global issues. Experimental results show the superior performance of our approach in terms of visual quality. [Copyright &y& Elsevier]

Abstract: The appearance of a face will vary drastically when the illumination changes. Variations in lighting conditions make face recognition an even more challenging and difficult task. In this paper, we propose a novel approach to handle the illumination problem. Our method can restore a face image captured under arbitrary lighting conditions to one with frontal illumination by using a ratio-image between the face image and a reference face image, both of which are blurred by a Gaussian filter. An iterative algorithm is then used to update the reference image, which is reconstructed from the restored image by means of principal component analysis (PCA), in order to obtain a visually better restored image. Image processing techniques are also used to improve the quality of the restored image. To evaluate the performance of our algorithm, restored images with frontal illumination are used for face recognition by means of PCA. Experimental results demonstrate that face recognition using our method can achieve a higher recognition rate based on the Yale B database and the Yale database. Our algorithm has several advantages over other previous algorithms: (1) it does not need to estimate the face surface normals and the light source directions, (2) it does not need many images captured under different lighting conditions for each person, nor a set of bootstrap images that includes many images with different illuminations, and (3) it does not need to detect accurate positions of some facial feature points or to warp the image for alignment, etc. [Copyright &y& Elsevier]

High dynamic range (HDR) imaging is an important task in image processing that aims to generate well-exposed images in scenes with varying illumination. Although existing multi-exposure fusion methods have achieved impressive results, generating high-quality HDR images in dynamic scenes remains difficult. The primary challenges are ghosting artifacts caused by object motion between low dynamic range images and distorted content in underexposure and overexposed regions. In this paper, we propose a deep progressive feature aggregation network for improving HDR imaging quality in dynamic scenes. To address the issues of object motion, our method implicitly samples high-correspondence features and aggregates them in a coarse-to-fine manner for alignment. In addition, our method adopts a densely connected network structure based on the discrete wavelet transform, which aims to decompose the input features into multiple frequency subbands and adaptively restore corrupted contents. Experiments show that our proposed method can achieve state-of-the-art performance under different scenes, compared to other promising HDR imaging methods. Specifically, the HDR images generated by our method contain cleaner and more detailed content, with fewer distortions, leading to better visual quality. [ABSTRACT FROM AUTHOR]

Saliency detection has extensive applications in daily life. In this paper, an efficient saliency-detection method based on wavelet transform and entropy theory is proposed. In the algorithm proposed in this paper, salient regions are viewed as uncommon regions in the background of an image. The uncommon regions can be caused by differences in color, orientation, texture, shape, or other factors. Considering the fact that wavelet coefficients can represent the local features of an image in different scales and orientations, the wavelet transform is therefore employed to identify the salient regions. Unlike those conventional wavelet-based methods, our proposed method need not perform the inverse wavelet transformation; this can reduce the computational requirements. In addition, because the different factors (i.e. color, orientation, texture, shape, etc.) stimulate different aspects of the human visual system, a saliency-map combination scheme based on the entropy theory is devised in this paper, which can evaluate the influence or significance of the different factors. Experimental results show that our method, based on wavelet transformation and entropy theory, can achieve excellent performance in terms of the area under the receiver operating characteristic curve (AUC) score, the linear correlation coefficient (CC), the normalized scan-path saliency (NSS) score, and visual performance, as compared to existing state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Most existing person re-identification (ReID) methods rely only on the spatial appearance information from either one or multiple person images, whilst ignore the space-time cues readily available in video or image-sequence data. Moreover, they often assume the availability of exhaustively labelled cross-view pairwise data for every camera pair, making them non-scalable to ReID applications in real-world large scale camera networks. In this work, we introduce a novel video based person ReID method capable of accurately matching people across views from arbitrary unaligned image-sequences without any labelled pairwise data. Specifically, we introduce a new space-time person representation by encoding multiple granularities of spatio-temporal dynamics in form of time series. Moreover, a Time Shift Dynamic Time Warping (TS-DTW) model is derived for performing automatically alignment whilst achieving data selection and matching between inherently inaccurate and incomplete sequences in a unified way. We further extend the TS-DTW model for accommodating multiple feature-sequences of an image-sequence in order to fuse information from different descriptions. Crucially, this model does not require pairwise labelled training data (i.e. unsupervised) therefore readily scalable to large scale camera networks of arbitrary camera pairs without the need for exhaustive data annotation for every camera pair. We show the effectiveness and advantages of the proposed method by extensive comparisons with related state-of-the-art approaches using two benchmarking ReID datasets, PRID2011 and iLIDS-VID. [ABSTRACT FROM AUTHOR]