Fine-grained Texture Research Articles

Increasingly Specialized Generative Adversarial Network for fine-grained visual categorization

Fine-grained visual categorization is challenging because the subordinate categories within an entry-level category can only be distinguished by subtle discriminations. This necessitates to localize key (most discriminative) regions and extract domain-specific features alternately, since implicit to fine-grained specialization is the existence of an entry-category visual shared among all classes. Existing methods predominantly implement fine-grained categorization independently, while neglecting that patch proposal and discrimination extraction are mutually correlated and can reinforce each other in an increasingly specialized manner. In this work, we concretize the above pipeline as an Increasing Specialized Generative Adversarial Network (IS-GAN), which recursively shapes a coarse-to-fine representation. It is a three-scale framework consisting of two highlights: a three-player expert GAN at each scale for feature extraction, and a Patch Proposal Network (PPN) between two adjacent scales for target positioning. To better anatomize pixel-to-pixel correlations at various octaves, the Gaussian pyramid and Laplacian pyramid descriptions are also integrated in each GAN. The PPN zooms the areas to shift the focus on the most representative regions by taking previous prediction of classifier as a reference, whilst a finer scale network receives an amplified attended region from previous scale. Overall, IS-GAN is driven by three focal losses from GANs and a converged object-level loss. Experiments demonstrate that IS-GAN can simultaneously (1) deliver competitive categorization performance among state-of-the-arts, i.e., validation accuracy achieves 92.23% and testing accuracy achieves 90.27%, and (2) recover fine-grained textures with high Peak Signal-to-Noise Ratios (PSNRs) (32.937) and Structural Similarities (SSIMs) (0.8607) from hand-crafted and public benchmarks.

Changes of phosphorus delivery from Yangtze River to Dongting Lake under new water and sediment conditions

Impounding the Three Gorges Reservoir (TGR) has altered the water-sediment transportation regime in the Yangtze River (YR) downstream, with inevitable effects on phosphorus (P) transport behavior in river-connected lakes. In this study, we investigate the characteristics of water, suspended sediment (SS), and P transportation, adsorption, and bioavailability at the three inlets along YR diverted to Dongting Lake through field surveys, adsorption experiments, and historical data analysis. The results showed that P transportation was significantly correlated with sediment (p < 0.01) and fine sediment particles dominated the transportation of particulate phosphorus (PP) and bioavailable phosphorus (Bio-P). Before the TGR impoundment, the dissolved inorganic phosphorus (DIP) concentration (0.03 mg/L) at the three inlets maintained a low level despite high P input, which was mainly due to the P adsorption and removal effect by sediment with high concentration and fine-grained texture. In addition, we reveal two changes at the three inlets caused by the impoundment: (1) SS content and discharge were reduced by an order of magnitude and the particle size was coarsened by three times; (2) the PP flux and bioavailability significantly decreased, whereas the transportation proportion of DIP increased. Consequently, the P buffering and adjusting capacity by sediment in the pristine river ecosystem from dam upstream supplies and downstream resuspension were notably weakened. Currently, the SS content at the three inlets (0.07 g/L) is far below the lower critical threshold (0.20 g/L) obtained through experiments that can effectively remove phosphate. Additionally, the decline of water and sediment supply from YR to Dongting Lake increased water transparency and prolonged water retention time in lake. With continuous TGR regulation and construction of upstream cascade reservoirs, the risk of algal blooms in river-connected lakes is significantly exacerbated.

Efficient texture-aware multi-GAN for image inpainting

Recent GAN-based (Generative adversarial networks) inpainting methods show remarkable improvements and generate plausible images using multi-stage networks or Contextual Attention Modules (CAM). However, these techniques increase the model complexity limiting their application in low-resource environments. Furthermore, they fail in generating high-resolution images with realistic texture details due to the GAN stability problem. Motivated by these observations, we propose a multi-GAN architecture improving both the performance and rendering efficiency. Our training schema optimizes the parameters of four progressive efficient generators and discriminators in an end-to-end manner. Filling in low-resolution images is less challenging for GANs due to the small dimensional space. Meanwhile, it guides higher resolution generators to learn the global structure consistency of the image. To constrain the inpainting task and ensure fine-grained textures, we adopt an LBP-based loss function to minimize the difference between the generated and the ground truth textures. We conduct our experiments on Places2 and CelebHQ datasets. Qualitative and quantitative results show that the proposed method not only performs favorably against state-of-the-art algorithms but also speeds up the inference time.

Feature-level interpolation-based GAN for image super-resolution

Image super-resolution is widely applied in face recognition, video perception, medical imaging, and many other fields. Although significant progress has been made, existing methods remain limited in reconstructing fine-grained texture details, making the pixels of the resulting images coarse. To address this problem, we propose a novel interpolation-based generative adversarial network (GAN) for high-resolution image reconstruction. First, an interpolation algorithm is introduced into the generator to carry out self-interpolation and channel interpolation using advanced features extracted from the low-resolution images. Second, the idea of residuals is introduced into both the generator and discriminator to expand the receptive field of the model and fully exploit the global features of the image jointly improving the visual perception of the resulting super-resolution image. Extensive experiments are conducted to evaluate the performances of the proposed models from two aspects: convergence speed and the resolution improvement effect. The experimental results demonstrate that the proposed model reaches a faster convergence speed with and comparable resolution improvement effect with respect to other state-of-the-art methods.

Semantic-SCA: Semantic Structure Image Inpainting With the Spatial-Channel Attention

Deep learning has brought unprecedented progress to image inpainting. However, the existing methods often generate images with blurry textures and distorted structures because they may either fail to maintain semantic consistency or restore fine-grained textures. In this paper, we propose a two-stage adversarial model to further improve the accuracy of the structure and details of image inpainting. Our model splits the inpainting task into two parts: semantic structure reconstructor and texture generator. In the first stage, we first utilize the semantic structure map based on the unsupervised segmentation to train the semantic structure reconstructor, which completes the missing structures of the inputs and maintains consistency between the missing part and the overall image. In the second stage, we introduce the spatial-channel attention (SCA) module to obtain the fine-grained textures. The SCA module strengthens the capability to obtain information from the long-distance pixel and different channels of the model. Furthermore, we propose a spatial-channel loss to stabilize the network training process and improve visual effects. Finally, we evaluate our model over the publicly available datasets CelebA, Places2, and Paris StreetView. When the inpainting tasks involved in large-area defects or heavy structure, the experimental results show that our method has a higher inpainting quality than the existing state-of-the-art approaches.

Effective Defect Detection Method Based on Bilinear Texture Features for LGPs

Automatic defect detection of light guide plates (LGPs) is an important task in the manufacture of liquid crystal displays. During thermo-printing, defects of tag lines on LGPs may occur easily, and these defects are of two categories: bubbles and missing tag lines. These defects lack salient visual attributes, such as edge-based and region-based features, and as such, traditional methods fail to detect them. To address this, we propose a Dense-bilinear convolutional neural network (BCNN), an end-to-end defect detection network, utilizing Dense-blocks (Huang et al. , 2017), Bilinear feature layers (Lin et al. , 2015), and squeeze-and-excitation blocks (Hu et al. , 2018). Our network exploits fine-grained texture features, which leads to parameter reduction and accuracy enhancement. We validate our network on our LGP dataset containing 5,860 images from three cases: bubbles, tag line existence, and tag line missing. Our network outperforms AlexNet (Krizhevsky et al. , 2012), VGG (Simonyan and Zisserman, 2014) and ResNet (He et al. , 2016), on both the public and our LGP datasets with less GPU memory consumption.

Global-Affine and Local-Specific Generative Adversarial Network for semantic-guided image generation

&lt;p style='text-indent:20px;'&gt;The recent progress in learning image feature representations has opened the way for tasks such as label-to-image or text-to-image synthesis. However, one particular challenge widely observed in existing methods is the difficulty of synthesizing fine-grained textures and small-scale instances. In this paper, we propose a novel Global-Affine and Local-Specific Generative Adversarial Network (GALS-GAN) to explicitly construct global semantic layouts and learn distinct instance-level features. To achieve this, we adopt the graph convolutional network to calculate the instance locations and spatial relationships from scene graphs, which allows our model to obtain the high-fidelity semantic layouts. Also, a local-specific generator, where we introduce the feature filtering mechanism to separately learn semantic maps for different categories, is utilized to disentangle and generate specific visual features. Moreover, we especially apply a weight map predictor to better combine the global and local pathways considering the highly complementary between these two generation sub-networks. Extensive experiments on the COCO-Stuff and Visual Genome datasets demonstrate the superior generation performance of our model against previous methods, our approach is more capable of capturing photo-realistic local characteristics and rendering small-sized entities with more details.&lt;/p&gt;

Deep-Learning-Based Gridded Downscaling of Surface Meteorological Variables in Complex Terrain. Part II: Daily Precipitation

AbstractStatistical downscaling (SD) derives localized information from larger-scale numerical models. Convolutional neural networks (CNNs) have learning and generalization abilities that can enhance the downscaling of gridded data (Part I of this study experimented with 2-m temperature). In this research, we adapt a semantic-segmentation CNN, called UNet, to the downscaling of daily precipitation in western North America, from the low resolution (LR) of 0.25° to the high resolution (HR) of 4-km grid spacings. We select LR precipitation, HR precipitation climatology, and elevation as inputs; train UNet over the subset of the south- and central-western United States using Parameter–Elevation Regressions on Independent Slopes Model (PRISM) data from 2015 to 2018, and test it independently in all available domains from 2018 to 2019. We proposed an improved version of UNet, which we call Nest-UNet, by adding deep-layer aggregation and nested skip connections. Both the original UNet and Nest-UNet show generalization ability across different regions and outperform the SD baseline (bias-correction spatial disaggregation), with lower downscaling error and more accurate fine-grained textures. Nest-UNet also shares the highest amount of information with station observations and PRISM, indicating good ability to reduce the uncertainty of HR downscaling targets.

Controllable Multi-Attribute Editing of High-Resolution Face Images

In recent years, significant progress has been achieved in face image editing due to the success of Generative Adversarial Network (GAN). However, state-of-the-art face editing methods mainly suffer from the following two limitations: 1) they are only applicable to face images with relative low-resolutions and 2) multi-attribute face editing may generate uncontrollable changes in non-target face attribute categories. To solve these problems, we propose a novel High-Quality Generative Adversarial Network (HQ-GAN) for controllable editing of multiple face attributes in high-resolution images. HQ-GAN has two novel ideas to break the limitations of resolution and controllability correspondingly: 1) fine-grained textures and realistic details of high-resolution face images are better preserved with the aid of textural features extracted by the wavelet transform module and 2) desired multi-attribute targets of face editing are emphasized using a weighted binary cross-entropy (BCE) loss so that the influence on non-target attributes is greatly reduced. To the best of our knowledge, HQ-GAN is the first attempt to achieve continuous editing of multiple face attributes on high-resolution images of the CelebA-HQ using only 28 000 training samples. Extensive qualitative results demonstrate the superiority of the proposed method in rendering realistic high-resolution face images with accurate attribute modification, and comprehensive quantitative results show that the proposed method significantly outperforms state-of-the-art face editing methods.

Image super‐resolution based on conditional generative adversarial network

Generative adversarial network (GAN) is one of the most prevalent generative models that can synthesise realistic high-frequency details. However, a mismatch between the input and the output may arise when GAN is directly applied to image super-resolution. To alleviate this issue, the authors adopted a conditional GAN (cGAN) in this study. The cGAN discriminator attempted to guess whether the unknown high-resolution (HR) image was produced by the generator with the aid of the original low-resolution (LR) image. They propose a novel discriminator that only penalises at the scale of the patch and, thus, has relatively few parameters to train. The generator of cGAN is an encoder–decoder with skip connections to shuttle the shared low-level information directly across the network. To better maintain the low-frequency information and recover the high-frequency information, they designed a generator loss function combining adversarial loss term and L1 loss term. The former term is beneficial to the synthesis of fine-grained textures, while the latter is responsible for learning the overall structure of the LR input. The experiments revealed that the proposed method could generate HR images with richer details and less over-smoothness.

A deep attention-based ensemble network for real-time face hallucination

Face hallucination (FH) aims to reconstruct high-resolution faces from low-resolution face inputs, making it significant to other face-related tasks. Different from general super resolution issue, it often requires facial priors other than general extracted features thus leading to fusion of more than one kind of feature. The existing CNN-based FH methods often fuse different features indiscriminately which may introduce noises. Also the latent relations among different features which may be useful are taken into less consideration. To address the above issues, we propose an end-to-end deep ensemble network which aggregates three extraction sub-nets in attention-based manner. In our ensemble strategy, both relations among different features and inter-dependencies among different channels are dug out through the exploitation of spatial attention and channel attention. And for the diversity of extracted features, we aggregate three different sub-nets, which are the basic sub-net for basic features, the auto-encoder sub-net for facial shape priors and the dense residual attention sub-net for fine-grained texture features. Conducted ablation studies and experimental results show that our method achieves effectiveness not only in PSNR (Peak Signal to Noise Ratio) and SSIM (Structural Similarity Index) metrics but more importantly in clearer details within both key facial areas and whole range. Also results show that our method achieves real-time hallucinating faces by generating one image in 0.0237s.

Petrographic Study and Geochemical Analysis of Basement Rocks in Federal University Gusau and Its Environs, Zamfara State, Northwestern Nigeria

Petrographic and geochemical analysis of basement rocks in Federal University Gusau and its environs, Zamfara State, Northwestern Nigeria were carried out in order to unveil the major rock forming minerals as well as the textural relationships composition within the rock. The identified rock unit is granite and they are mainly whitishgrey in colour, with the majority of them having porphyritic-coarse-grained texture and minor with fine-grained texture. The light coloured crystals in hand specimens are the felsic minerals mainly quartz, and feldspars while the grey-dark crystals are the mafic minerals like the biotite-mica. The result of the thin section analysis depicts that the minerals present within the rock samples in the area: quartz (40%), orthoclase (40%) and biotite (20%). These identified minerals are classified as the major rock forming silicate minerals. The geochemical analysis result shows that the rocks are characterized by high percentage of quartz (SiO2) in the range of 77.4-98.2 wt% with an average of 84.87 wt%; subordinate amounts of hematite (Fe2O3) with an average concentration of 4.08 wt% as well as low concentration of K2O, CaO, TiO2, V2O5, MnO, NiO, CuO, ZnO, RuO2, In2O3 and other minor constituent oxides. The variation diagram of SiO2 and other oxides within the study area reveals a very strong negative correlation of SiO2 against CaO, TiO2, V2O5, MnO, Fe2O3, CUO, PbO, and Eu2O3 respectively, with the average correlation value of -0.91. The geochemical analysis results also reveal minor alteration of modal composition amongst the samples analysed which suggest the similarities in composition of all the samples within the study area.

Contribution to the origin of Mn-U-Be-HREE-enrichment in phosphorite, near Bükkszentkereszt, NE Hungary

Strongly deformed phosphorite layers enriched with U-Be-Mn-REE occur in a weathered Triassic metarhyolite tuff near the town of Bükkszentkereszt, NE Hungary. The phosphorite is massive or earthy, has an especially fine-grained texture, and forms inhomogeneous bands with different Mn-oxide contents.High-resolution optical rock microscopy investigations helped to distinguish mineralized biosignatures, filamentous, vermiform structures, and coccoid-like aggregates. These microstructures encompass almost the whole phosphorite. Based on in situ FTIR and Raman spectroscopy, the phosphorites contain ferrihydrite and pyrite, and also different types of embedded organic matter. These structures can be interpreted as series of Fe-rich biomats, forming microbialites. On a larger scale, these microstructures can play a role in shaping the stratiform structure by evolving stromatolite-like bands.Results presented here propose a new interpretation of the origin of the P-Mn-U-Be-HREE-enrichment. The Bükkszentkereszt occurrence claim a mineralized microbially mediated deposit, and our results support this scenario based on the mineralized microbial structures, the embedded minerals, and the presence of organic matter.

Geochemistry of Garga-Sarali intrusive granitoids (central domain of the central African fold belt in Cameroon): petrological implication

The Garga-Sarali granitoids outcrop in form of large slabs and undistorted large blocks, into a schisto-gneissic basement. These rocks contain mainly muscovite and microcline, followed by K-feldspar, quartz, biotite, pyroxene, zircon and oxides, with coarse-grained to fine-grained textures. Geochemical analysis show that it belongs to differentiated rocks group (granodiorite-granite) with high SiO2 (up to 72 wt%) contents. Their genesis was made from a process of partial melting and fractional crystallization. These rocks are classified as belonging to I- and S-Type, meta-peraluminous, shoshonitic granites; belonging to the domain of volcanic arcs. The rare earth elements patterns suggest a source enriched of incompatible elements. The Nb-Ta and Ti negative anomalies from the multi-element patterns are characteristics of the subduction domains.

A novel quadruple generative adversarial network for semi-supervised categorization of low-resolution images

In order to make utilization of unlabeled low-resolution (LR) images to shape discriminative models, we present quadruple generative adversarial network (Q-GAN), a game-theoretical framework for implementing semi-supervised categorization of LR images. It can realize photo-realistic image super-resolution (SR) and semi-supervised pattern recognition simultaneously. We consider our pipeline as a four-player optimization-based formulation, which consists of four vital components, i.e., a refiner for image SR and generation, a discriminator for identifying high-resolution (HR) samples and another for identifying true (original) samples, a classifier for label prediction. The refiner and two discriminators characterize the conditional distributions between images and labels, whilst the classifier solely focuses on predicting real image-label pairs. We select those high-quality super-solved images with ground-truth labels for data supplement. We customize the global optimization objective function as well as the training procedure to ensure model approximates the posterior distribution of latent variables given true data in a semi-supervised manner. Experimental results demonstrate that Q-GAN can simultaneously (1) deliver the promising categorization performance among state-of-the-arts, i.e., validation accuracy achieves 92.18% and testing accuracy achieves 90.63%, and (2) recover fine-grained textures with high peak signal-to-noise ratios (PNSRs) and structural similarities (SSIMs) from heavily downsampled testing images of hand-crafted dataset and public benchmarks.

Fabric surface characterization: assessment of deep learning-based texture representations using a challenging dataset

Tactile sensing or fabric hand plays a critical role in an individual’s decision to buy a certain fabric from the range of available fabrics for a desired application. Therefore, textile and clothing manufacturers have long been in search of an objective method for assessing fabric hand, which can then be used to engineer fabrics with a desired hand. In this paper, we explore how to characterize surface properties (e.g. smoothness) of materials. We formulate the problem as a fine-grained texture classification problem, and study how deep learning-based texture representation techniques can help tackle the task. We introduce a new, challenging microscopic material surface dataset (CoMMonS), geared towards an automated fabric quality assessment mechanism in an intelligent manufacturing system. Additionally, we propose a multi-level texture encoding and representation network (MuLTER), which extracts texture details and structural information. Our dataset and source code are available at https://ghassanalregib.info/software-and-datasets.

Fingerprint Liveness Detection Based on Fine-Grained Feature Fusion for Intelligent Devices

Currently, intelligent devices with fingerprint identification are widely deployed in our daily life. However, they are vulnerable to attack by fake fingerprints made of special materials. To elevate the security of these intelligent devices, many fingerprint liveness detection (FLD) algorithms have been explored. In this paper, we propose a novel detection structure to discriminate genuine or fake fingerprints. First, to describe the subtle differences between them and take advantage of texture descriptors, three types of different fine-grained texture feature extraction algorithms are used. Next, we develop a feature fusion rule, including five operations, to better integrate the above features. Finally, those fused features are fed into a support vector machine (SVM) classifier for subsequent classification. Data analysis on three standard fingerprint datasets indicates that the performance of our method outperforms other FLD methods proposed in recent literature. Moreover, data analysis results of blind materials are also reported.

Parauapebas meteorite from Pará, Brazil, a “hammer” breccia chondrite

Abstract The Parauapebas meteorite, third official meteorite discovered in the Brazilian Amazon region, is a “hammer meteorite” which fell on December 9th, 2013, in the city of Parauapebas, Para State, Brazil. Mineralogy is dominated by forsterite, enstatite, iron, troilite, and tetrataenite. Albite, chromite, diopside, augite, pigeonite, taenite, and merrillite are minor components. Two main clasts are separated by black shock-induced melt veins. One clast exhibits an abundance of chondrules with well-defined margins set on a recrystallized matrix composed mostly of forsterite and enstatite, consistent with petrologic type 4 chondrites. The other clast displays chondrules with outlines blurring into the groundmass as evidence of increasing recrystallization, consistent with petrologic type 5 chondrites. The clasts of petrologic type 4 have a fine-grained texture compared to those of type 5. It is a genomict breccia (indicated by shock melt veins) with the clasts and matrix of the same compositional group, but different petrologic types, H4 and H5. The melted outer crust of the Parauapebas meteorite is comprised of forsterite with interstitial dendritic iron oxide, and is rich in irregular vesicles, which are evidence of the rapid formation of the crust. The type specimen is deposited in the Museum of Geosciences of the University of Sao Paulo, Brazil.

Segmentation of Visual Images by Sequential Extracting Homogeneous Texture Areas

The purpose of the research is to develop a universal algorithm for partial texture segmentation of any visual images. The main peculiarity of the proposed segmentation procedure is the extraction of only homogeneous fine-grained texture segments present in the images. At first, an initial seed point is found for the largest and most homogeneous segment of the image. This initial seed point of the segment is expanded using a region growing method. Other texture segments of the image are extracted analogously in turn. At the second stage, the procedure of merging the extracted segments belonging to the same texture class is performed. Then, the detected texture segments are input to a neural network with competitive layers which accomplishes more accurate delineation of the shapes of the extracted texture segments. The proposed segmentation procedure is fully unsupervised, i.e., it does not use any a priori knowledge on either the type of textures or the number of texture segments in the image. The research results in development of the segmentation algorithm realized as a computer program tested in a series of experiments that demonstrate its efficiency on grayscale natural scenes.

Synthesis and characterization of NaAlSi2O6 jadeite doped with different metallic oxides under high pressure and high temperature

According to the mineral composition [Formula: see text] of jadeite, [Formula: see text] and [Formula: see text] with a molar ratio of 1:1 were selected as raw materials. One sample was undoped, and five samples were mixed with 0.4 wt.% [Formula: see text], MnO, CoO, [Formula: see text] and [Formula: see text], respectively. The experimental studies were executed under the synthetic condition of 5 GPa pressure and [Formula: see text] temperature using China-type large volume cubic high-pressure apparatus (CHPA) (SPD-6X1200). After the experiment, compositions of the synthetic jadeites were characterized by X-ray diffraction, microstructures characterized by field emission scanning electron microscopy (FESEM) and the high-resolution transmission electron microscopy (HRTEM), molecular vibration types investigated by Raman spectra. The color-causing ion of the synthetic jadeites was investigated by UV–Vis Spectra. The results show that the synthetic jadeites have excellent crystallinity and fine-grained texture and similar structural behavior with natural jadeite. UV–Vis Spectra indicated that different colors of samples have different ion absorption peaks, thus showing different colors. This experiment explored the geological conditions of synthetic jadeite by means of high pressure and high temperature (HPHT) to give a favorable experimental basis for synthetic jadeite and analyze the coloration mechanism of jadeite by way of ion-doping.