Color Discrepancy Research Articles

Development of anti-bacterial bio-transfer double sheet layer of polyvinyl alcohol/carboxymethyl cellulose films infused with Astragalus tribuloides leaf extract for beef burgers preservation

This study was conducted to develop biodegradable films using a combination of carboxymethyl cellulose (CMC), polyvinyl alcohol (PVA) and purified leaves extract of Astragalus tribuloides (ATE). Various traits of the films, including their morphology description, thermal behavior, tensile/elongation properties and physical characteristics were examined. The scanning electron microscope (SEM) photographs showed smooth surface with small amounts of ATE, but rougher with higher concentrations of 1.4 %. The Fourier-transform infrared spectroscopy (FTIR) showed a direct relationship between the ATE extract and the PVA/CMC matrix. The films also showed thermal stability behaviors. The study found that the addition of ATE up to 0.8 % caused the films to become opaquer in color and raised their opacity up to 3.909. As a result, the films exhibited reduced moisture absorption (8.21 %) and solubility (27.11 %), making them retard penetrating water vapor up to (1.785 × 10−10 g.m−1 s-1 Pa−1) and could preserve the thiobarbituric acid reactive substances (TBARS) and overall color discrepancies of burger in refrigerated storage.

Optimized single-bath dyeing of bio-based nylon 56/cotton blends with vat dyes: Mechanistic insights and process enhancements

Nylon 56/cotton (N/C) blended fabrics have excellent durability and can cater to various clothing needs. However, the quality and value of textile garments are significantly influenced by the dyeing process. Due to variations in the structure and characteristics of nylon 56 and cotton fibers in blended fabrics, achieving consistent coloration is challenging, resulting in noticeable color discrepancies. To tackle these issues, this study focused on dyeing nylon 56/cotton blended fabrics in a single bath using vat dyes. The three primary colors of the vat dyes were selected to be suitable for these fabrics. Subsequently, the study explored the impact of different factors on color consistency and optimized the dyeing process conditions using response surface methodology. The findings revealed that fabrics dyed using the optimized process exhibited vibrant colors, uniformity, and excellent colorfastness to soaping and rubbing, surpassing standard requirements. Furthermore, the study delved into the dyeing mechanism of nylon 56/cotton blended fabrics, analyzing the kinetic and thermodynamic properties of dye adsorption on fibers. This analysis aims to provide a theoretical foundation for developing a rational dyeing process and offer guidance for practical production.

A Clinical Comparison Between New and Conventional Methods for Quantification of Intended and Fabricated Dental Porcelain Color

Introduction: The accurate selection of dental porcelain shades is crucial for achieving natural-looking and aesthetically pleasing restorations. This study aimed to compare the effectiveness of various shade selection methods, including visual assessment, spectrophotometry, and intraoral scanning. Aims and Objectives: This study aims to improve the accuracy of shade selection in prosthodontics. To achieve this, it will compare digital, intraoral, and visual methods, identify factors influencing shade selection, develop educational guidelines for dentists, and minimize color discrepancies between chosen and final shades. Methods: A diverse group of 37 participants with 111 shades of teeth were involved. Three shade selection methods were employed: visual assessment using a shade guide, spectrophotometry with the VITA Easyshade Compact, and intraoral scanning with the Medit i700. The accuracy of each method was determined by comparing the selected shade to the fabricated porcelain restoration. Results: While digital methods (spectrophotometry and intraoral scanning) simplified the shade selection process, visual assessment, especially when performed by experienced dentists, consistently demonstrated higher accuracy. Factors such as lighting conditions and individual variations in color perception influenced the reliability of visual assessment. Discussion: The study emphasizes the importance of a comprehensive approach to shade selection, combining both visual and digital methods. Digital tools offer objective measurements, while visual assessment captures subtle nuances in color and ensures patient satisfaction. Conclusion: The optimal shade selection method depends on the specific clinical situation and the dentist's expertise. By incorporating both visual and digital techniques, dental professionals can enhance the accuracy and reliability of shade selection, ultimately leading to more natural-looking and aesthetically pleasing dental restorations.

Effect of Hydrogen Peroxide-Based Bleaching Agents on the Color Dimensions and Surface Roughness of Different Milled Restorative Dental Materials.

To compare the color dimensions, color discrepancies (ΔE00), and surface roughness of milled materials before and after application of a bleaching agent. A total of 10 extracted molars were obtained. Each tooth was cut in transverse sections to create disks (3-mm thick, 10-mm diameter; control group). Disk specimens of eight materials (n &#61; 10 per group) were fabricated: polymethyl methacrylate (PMMA) interim material (PMMA-Telio group), two resin nanoceramics (RNC-Ultimate group and RNC-Cerasmart group), two hybrid ceramics (HC-Shofu group and HC-Enamic group), lithium disilicate (LD-Emax group), zirconia-reinforced glass ceramic (ZGC-Suprinity group), and zirconia (Zr-InCeram group). Color measurements were obtained using a spectrophotometer before and after applying 35% hydrogen peroxide-based bleaching agent. Pre- and postbleaching surface roughness (Sa) analyses were completed using a profilometer. Significant L*, a*, b*, and ΔE00 value differences were found (P < .05). Color discrepancies (ΔE00) ranged from 0.30 ± 0.14 to 4.82 ± 0.10. The highest color discrepancies were measured on the PMMA-Telio group, while the lowest color discrepancies were computed for ZGC-Suprinity, RNCUltimate, and RNC-Cerasmart. Significant surface roughness differences were found (P < .05). The largest increase of surface roughness values between the pre- and postbleaching measurements was obtained in the PMMA-Telio group with a mean ΔSa value of 4.73 ± 3.02, while the largest decrease of surface roughness values between the pre- and postbleaching measurements was obtained in the Zr-InCeram group with a mean ΔSa value of -1.58 ± 0.10. The milled materials showed significant pre- and postbleaching color and surface roughness discrepancies.

A Texture-Considerate Convolutional Neural Network Approach for Color Consistency in Remote Sensing Imagery

Remote sensing allows us to conduct large-scale scientific studies that require extensive mapping and the amalgamation of numerous images. However, owing to variations in radiation, atmospheric conditions, sensor perspectives, and land cover, significant color discrepancies often arise between different images, necessitating color consistency adjustments for effective image mosaicking and applications. Existing methods for color consistency adjustment in remote sensing images struggle with complex one-to-many nonlinear color-mapping relationships, often resulting in texture distortions. To address these challenges, this study proposes a convolutional neural network-based color consistency method for remote sensing cartography that considers both global and local color mapping and texture mapping constrained by the source domain. This method effectively handles complex color-mapping relationships while minimizing texture distortions in the target image. Comparative experiments on remote sensing images from different times, sensors, and resolutions demonstrated that our method achieved superior color consistency, preserved fine texture details, and provided visually appealing outcomes, assisting in generating large-area data products.

Illumination correction for dyed products with spotted hyena optimizer and regularized extreme learning machine

Dyed products exhibit varying colors when illuminated by different light sources. Human eyes possess color constancy, enabling them to automatically adjust and overlook color discrepancies, effectively self-correcting visually. However, these variations are perceived as distinct colors by computers. To eliminate the effects of lighting on the dyed products, illumination correction is needed to make the dyed products show their original color. To achieve this goal, this article presents a novel algorithm, WOA-SHO-RELW (Whale Optimization Algorithm-Spotted Hyena Optimizer-Regularized Extreme Learning Machine), designed to emulate color constancy and applies it to the illumination correction of dyed products. We employ the WOA to optimize the initial population of SHO, carefully selecting a group of suitable initial parameters. Subsequently, SHO is utilized to optimize the hidden biases and input weights of RELM, with the overarching aim of enhancing the model’s effectiveness and performance, thereby ensuring more precise illumination correction for images of dyed products. The comparison results with other algorithms show that our WOA-SHO-RELM has better comprehensive performances.

Assessment of Variability in Free Flap Color Match to Facial Skin by Donor Site and Race.

To use portable colorimetry to quantify color differences between facial skin and potential three head and neck microvascular free tissue transfer (MFTT) donor sites-radial forearm (RF), anterolateral thigh (ALT), and fibula (FF)-and compare these differences by pigmentation of the donor site skin and self-identified race. In this cross-sectional cohort study, healthy volunteers consented to handheld colorimeter measurements at the three potential MFTT donor sites (RF, ALT, FF) to quantify color match to the facial skin using the CIE color space (DeltaE). The comparison of ipsilateral to contralateral cheek served as control for measurements. Cross-sectional measurements in healthy volunteers were then compared to measurements obtained in postoperative head and neck MFTT patients. DeltaE measurements were obtained for 128 healthy controls and 24 postoperative patients (N = 152). With increasing lightness (decreased pigmentation) of the skin at the donor site, the color match significantly worsened (higher DeltaE) across all potential MFTT donor sites (all p < 0.05). DeltaE from healthy controls closely approximated postoperative color match measurements in patients who underwent cervicofacial MFTT (DeltaE RF: 5.3 vs. 6.0, p = 0.432; DeltaE ALT: 6.2 vs. 6.4, p = 0.822; DeltaE FF: 6.0 vs. 6.4, p = 0.806). Patients with decreased skin pigmentation who are undergoing head and neck MFTT may experience worse color discrepancy between cervicofacial skin and the transferred skin paddle than those with more pigmented skin. Portable colorimetry may identify patients who could benefit from interventions such as dermis-resected free tissue reconstruction with skin grafting to improve postoperative appearance. 3 Laryngoscope, 134:3581-3586, 2024.

Weighted Similarity-Confidence Laplacian Synthesis for High-Resolution Art Painting Completion

Artistic image completion assumes a significant role in the preservation and restoration of invaluable art paintings, marking notable advancements through the adoption of deep learning methodologies. Despite progress, challenges persist, particularly in achieving optimal results for high-resolution paintings. The intricacies of complex structures and textures in art paintings pose difficulties for sophisticated approaches like Generative Adversarial Networks (GANs), leading to issues such as small-scale texture synthesis and the inference of missing information, resulting in distortions in lines and unnatural colors. Simultaneously, patch-based image synthesis, augmented with global optimization on the image pyramid, has evolved to enhance structural coherence and details. However, gradient-based synthesis methods face obstacles related to directionality, inconsistency, and the computational burdens associated with solving the Poisson equation in non-integrable gradient fields. This paper introduces a pioneering approach, integrating Weighted Similarity-Confidence Laplacian Synthesis to comprehensively address these challenges and advance the field of artistic image completion. Experimental results affirm the effectiveness of our approach, offering promising outcomes for the preservation and restoration of art paintings with intricate details and irregular missing regions. The integration of weighted Laplacian synthesis and patch-based completion across multi-regions ensures precise and targeted completion, outperforming existing methods. A comparative analysis underscores our method’s superiority in artifact reduction and minimizing blurriness, particularly addressing challenges related to color discrepancies in texture areas. Additionally, the incorporation of pyramid blending proves advantageous, ensuring smoother transitions and preventing noticeable seams or artifacts in blended results. Based on empirical results, our method consistently outperforms previous methods across both high and low resolutions. Responding to these insights, our approach emerges as an invaluable guide for both curators and artists. The algorithm’s performance yields insights that underscore the central role of thoughtful decision making in the creation of art paintings. This guidance extends to informing choices related to color selection, brushstrokes, and various other elements integral to the artistic process. During the creation phase, employing these insights enables artists and curators to optimize not only the digitization but also the subsequent restoration process. This proves especially vital when dealing with the intricacies involved in physically restoring damaged original art paintings. Importantly, our approach not only streamlines the restoration process but also contributes significantly to the preservation and enhancement of the digital representations of these distinctive and often intricate works of art.

Development of a Machine Learning Model to Predict the Color of Extruded Thermoplastic Resins.

The conventional method for the color-matching process involves the compounding of polymers with pigments and then preparing plaques by using injection molding before measuring the color by an offline spectrophotometer. If the color fails to meet the L*, a*, and b* standards, the color-matching process must be repeated. In this study, the aim is to develop a machine learning model that is capable of predicting offline color using data from inline color measurements, thereby significantly reducing the time that is required for the color-matching process. The inline color data were measured using an inline process spectrophotometer, while the offline color data were measured using a bench-top spectrophotometer. The results showed that the Bagging with Decision Tree Regression and Random Forest Regression can predict the offline color data with aggregated color differences (dE) of 10.87 and 10.75. Compared to other machine learning methods, Bagging with Decision Tree Regression and Random Forest Regression excel due to their robustness, ability to handle nonlinear relationships, and provision of insights into feature importance. This study offers valuable guidance for achieving Bagging with Decision Tree Regression and Random Forest Regression to correlate inline and offline color data, potentially reducing time and material waste in color matching. Furthermore, it facilitates timely corrections in the event of color discrepancies being observed via inline measurements.

Optimized 3D reconstruction in nearshore underwater environments: A cost-effective pre-processing strategy for Neural Radiance Fields implementation

Nearshore oceans, teeming with diverse benthic ecosystems, continue to be a focal point for marine research. While 2D visual representations have been the mainstay in this area, the intricate, multi-dimensional nature of the seafloor ecosystems underscores the need for 3D modeling to capture their full essence. This study unveils a novel approach tailored for static image processing and 3D modeling through Neural Radiance Fields (NeRF), with a particular emphasis on the refined instantNGP variant. A meticulously crafted pipeline is employed, centered on neutralizing the visual impediments brought about by the interplay of light and seawater in underwater imaging. This refined pre-processing strategy ensures that images are primed for a seamless transition to NeRF-based 3D reconstruction, all the while conserving computational resources. The refined image processing techniques rectify underwater color discrepancies, notably the prevalent blue-green hue resulting from unique lighting conditions. Moreover, the system's ability to identify and excise seawater boundaries guarantees that the 3D models remain singularly focused on the richness of the seafloor ecosystems. Remarkably, achieving this does not demand vast datasets or exorbitant computational prowess, positioning it as an ideal fit for processing images from nearshore regions. As a more resource-friendly and efficient counterpart to existing methodologies, this study furnishes marine ecologists with a powerful instrument for RGB-centric 3D renderings of nearshore terrains. Nonetheless, for broader applicability in diverse marine settings, fusing this approach with neural networks could prove invaluable.

Color Change Analysis of Raw Chicken Breast over Different Timestamps Using CNNs

In this study, the potential of a novel method based on an artificial neural network was investigated to analyze the color change of raw chicken breasts during refrigeration. As edible meat for people, chicken breasts have high nutrients and low‐fat content. Therefore, people consume it as a safe and high‐value food in their daily diet. The investigation of chicken breast freshness is proposed as a significant issue in the meat industry because raw meat spills rapidly. The color change of raw chicken breasts over time reflects subtle biochemical changes, as well as changes in freshness. However, owing to the impact of the photographic equipment and illumination, a significant color discrepancy exists between the captured image and the intrinsic color. It is difficult to separate chicken breast color changes from color discrepancies when comparing images captured at different timestamps. Thus, we propose a color change analysis method for raw chicken breast that uses color correction to suppress the influence of color discrepancy and uses CNN to extract discriminant features for timestamp classification analysis. The experimental results indicate that the proposed method improves all three CNN models by achieving higher accuracy, and the best model was improved from to , demonstrating the effectiveness of the proposed method. © 2024 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC.

Color discrepancy of single-shade composites at different distances from the interface measured using cell phone images.

This study aimed to evaluate the impact of substrate color and interface distance on the color adjustment of 2 single-shade composites, Vittra APS Unique and Charisma Diamond One. Dual disc-shaped specimens were created using Vittra APS Unique or Charisma Diamond One as the center composite, surrounded by shaded composites (A1 or A3). Color measurements were taken with a spectrophotometer against a gray background, recording the color coordinates in the CIELAB color space. Illumination with a light-correcting device and image acquisition using a polarizing filter-equipped cell phone were performed on specimens over the same background. Image processing software was used to measure the color coordinates in the center and periphery of the inner composite and in the outer composite. The color data were then converted to CIELAB coordinates and adjusted using data from the spectrophotometer. Color differences (ΔE00) between the center/periphery of single-shade and outer composites were calculated, along with color changes in single-shade composites caused by different outer composites. Color differences for the inner composites surrounded by A1 and A3 were also calculated. Data were analyzed using repeated-measures analysis of variance (α = 0.05). The results showed that color discrepancies were lowest near the interface and when the outer composite was whiter (A1). Additionally, Charisma Diamond One exhibited better color adjustment ability than Vittra APS Unique. Color discrepancies between the investigated single-shade composites diminished towards the interface with the surrounding composite, particularly when the latter exhibited a lighter shade.

Surrounding and background shades effect on color adjustment of single-shade composites.

This study evaluated the effects of surrounding and background shades on the color adjustment potential (CAP) and visual color match of two single-shade composites, Vittra APS Unique and Charisma Diamond One. Cylinder-shaped specimens were constructed, consisting of either single-shade composites alone (simple) or composites surrounded by the Forma material at shades A1 or A3 (dual). Simple specimens using only the Forma at the specified shades were also prepared. Color measurements of simple specimens were taken against a gray background using a spectrophotometer. For dual specimens, the color of the inner composite was measured against a gray or chromatic (the same shade as the outer composite) background. Color differences (ΔE00) between the single-shade composites and the A1/A3 composite were calculated. CAP was determined by comparing data from simple and dual specimens. Four experienced dentists scored the color match (perfect to unacceptable) for each specimen using a viewing booth illuminated by an illuminant D65. Data were analyzed with repeated-measures ANOVA and the Kruskal-Wallis test. The results showed that both single-shade composites showed the lowest color discrepancies when they were compared to A1 and A3. Using a chromatic background only significantly affected the CAP when the outer composite was A3. The visual analysis showed poor color matches between the single-shade and control chromatic composites, except for Charisma Diamond One surrounded by A1. In conclusion, the CAP values of the evaluated single-shade composites were impacted by both surrounding shade and background color, and the color match of these materials tended to be poor.

Color matching and color recovery in large composite restorations using single-shade or universal composites.

This study assessed the color-matching ability and color recovery of unprepared teeth when using single-shade composites and a universal composite in large restorations. Buccal and palatine surface colors of molars were measured with a spectrophotometer (CIELAB) before preparing round cavities (6 mm in diameter, 2 mm in depth). The cavities were randomly filled with a single-shade composite (Omnichroma, Diamond One, or Vittra APS Unique) or a universal composite (Filtek Universal). Color measurements of the restored cavities were taken, and overall color differences (ΔEab and ΔE00) and differences in the whitening index for dentistry (ΔWID) from baseline were calculated. Additionally, visual assessments of a color match to the surrounding enamel were performed by forty evaluators (laypersons and undergraduate students of dentistry) in a viewing booth under illuminant D65, with rating scores from 0 (no color mismatch) to 4 (not acceptable). Data were analyzed using RM or one-way ANOVA (α = 0.05). Results showed that the restorations generally exhibited whiter colors (WID ranged from 27.9 to 41.3) than the unprepared teeth (WID ranged from 15.9 to 19.3). The composite Filtek Universal demonstrated the lowest color discrepancy (ΔWID = 8.6; ΔE00 = 10.8; and ΔE00 = 6.2), and no significant differences were observed among the evaluated single-shade composites. Furthermore, all composites showed similar and adequate color matches to the surrounding enamel. However, it is important to note that despite their ability to match the surrounding enamel reasonably, none of the composites evaluated in large restorations fully recovered the color observed in unprepared teeth.

Regionwise Generative Adversarial Image Inpainting for Large Missing Areas.

Recently, deep neural networks have achieved promising performance for in-filling large missing regions in image inpainting tasks. They have usually adopted the standard convolutional architecture over the corrupted image, leading to meaningless contents, such as color discrepancy, blur, and other artifacts. Moreover, most inpainting approaches cannot handle well the case of a large contiguous missing area. To address these problems, we propose a generic inpainting framework capable of handling incomplete images with both contiguous and discontiguous large missing areas. We pose this in an adversarial manner, deploying regionwise operations in both the generator and discriminator to separately handle the different types of regions, namely, existing regions and missing ones. Moreover, a correlation loss is introduced to capture the nonlocal correlations between different patches, and thus, guide the generator to obtain more information during inference. With the help of regionwise generative adversarial mechanism, our framework can restore semantically reasonable and visually realistic images for both discontiguous and contiguous large missing areas. Extensive experiments on three widely used datasets for image inpainting task have been conducted, and both qualitative and quantitative experimental results demonstrate that the proposed model significantly outperforms the state-of-the-art approaches, on the large contiguous and discontiguous missing areas.

Effect of hydrogen peroxide-based bleaching agents on the color dimensions and surface roughness of different milled restorative dental materials.

To compare the color dimensions, color discrepancies (ΔE00), and surface roughness of milled materials before and after the application of a bleaching agent. A total of 10 extracted molars were obtained. Each tooth was cut in transversal sections, to create discs (3 mm thick, 10 mm diameter) (control group). Disk specimens of 8 materials were fabricated: polymethyl methacrylate (PMMA) interim material (PMMA-Telio group), 2 resin nanoceramics: (RNC-Ultimate group) and (RNC-Cerasmart group), 2 hybrid ceramics: (HC-Shofu group) and (HC-Enamic group), lithium disilicate (LD-Emax group), zirconia reinforced glass ceramic (ZGC-Suprinity group), and zirconia (Zr group) (n=10). Color measurements were obtained using a spectrophotometer before and after applying 35% hydrogen peroxide-based bleaching agent. Pre- and post-bleaching surface roughness analysis were completed using a profilometer. Significant L*, a*, b*, and ΔE00 values differences were found (P < .05). Color discrepancies (ΔE00) ranged from 0.30 ± 0.14 to 4.82 ± 0.10. The highest color discrepancies were measured on the PMMA-Telio group, while the lowest color discrepancies were computed on the ZGC-Suprinity, RNC-Ultimate, and RNC-Cerasmart. Significant surface roughness differences were found (P <.05). The largest increase of surface roughness values between the pre- and post-bleaching measurements was obtained in the PMMA-Telio group with a mean ΔSa value of 4.73 ± 3.02, while the largest decrease of surface roughness values between the pre- and post-bleaching measurements was obtained in the Zr-InCeram group with a mean ΔSa value of -1.58 ± 0.10. The milled materials tested showed significant pre- and post-bleaching color and surface roughness discrepancies. Int J Prosthodont. doi: 10.11607/ijp.8359.

The peri-implant mucosa color: A systematic appraisal of methods for its assessment and clinical significance.

Peri-implant mucosa color (PMC) seems to be one of the main parameters affecting the esthetic outcome of implant therapy. However, more emphasis should be given to its assessment and reporting. To describe the available evidence on methods to assess and report the color of the peri-implant mucosa (PMC) and the respective clinical relevance. A comprehensive electronic and manual search was performed to identify clinical studies reporting on PMC. A total of 121 studies were included. PMC was evaluated at the time of the follow-up visit (chairside) in 45.5% studies. PMC assessment was performed qualitatively, by comparing PMC with adjacent and/or contralateral gingiva (78.6%) or quantitatively, using spectrophotometry (20.7%) or a software on clinical photographs (0.8%). The most performed method to assess PMC was through esthetic indices (76.9%), either at the time of the follow-up visit (chairside) or at later time point using photographs. Quantitative reporting of PMC included averages of points from esthetic indices or color differences to natural gingiva expressed with the CIELAB color system. PMC assessment allowed describing color discrepancies compared to natural gingiva, evaluating color changes over time, and comparing the outcomes of different treatment modalities. PMC assessment through spectrophotometry was additionally utilized to assess the role of mucosal thickness (MT) on PMC. Various methods for PMC assessment and reporting were described, including visual assessment, mainly through esthetic indices, and spectrophotometry. PMC evaluation has allowed to demonstrate the factors affecting the color of the peri-implant soft tissue, such as the type of abutment/restoration, MT, and soft tissue augmentation.

Structure-Texture Consistent Painting Completion for Artworks

Image completion techniques have made rapid and impressive progress due to advancements in deep learning and traditional patch-based approaches. The surrounding regions of a hole played a crucial role in repairing missing areas during the restoration process. However, large holes could result in suboptimal restoration outcomes due to complex textures causing significant changes in color gradations. As a result, they led to errors such as color discrepancies, blurriness, artifacts, and unnatural colors. Additionally, recent image completion approaches focused mainly on scenery and face images with fewer textures. Given these observations, we present a structure-texture consistent completion approach for filling large holes with detailed textures. Our method focuses on improving image completion in the context of artworks, which are expressions of creativity and often have more diverse structures and textures from applying paint to a surface using brush strokes. To handle the unique challenges posed by artwork, we segment non-homogeneous areas and then use Cohesive Laplacian Fusion to complete the texture of the remaining missing segmented area. This technique involves detecting changes in base structures and textures using multiple matched patches to achieve more consistent results. The experimental results show that our proposed method is competitive and outperforms state-of-the-art methods in missing regions and color gradations of art paintings.

Context Adaptive Network for Image Inpainting.

In a typical image inpainting task, the location and shape of the damaged or masked area is often random and irregular. The vanilla convolutions widely used in learning-based inpainting models treat all spatial features as valid and share parameters across regions, making it difficult for them to cope with those irregular damages, and models tend to produce inpainting results with color discrepancy and blurriness. In this paper, we propose a novel Context Adaptive Network (CANet) to address this issue. The main idea of the proposed CANet is able to generate different weights depending on the miscellaneous input, which may help to complement images with multiple broken forms in a flexible way. Specifically, the proposed CANet has two novel context adaptive modules, namely, the context adaptive block (CAB) and the cross-scale contextual attention (CSCA), which utilize attention mechanisms to cope with diverse content breakdowns. The proposed CAB, during the forward propagation, uses an adaptive term to determine the importance between adaptive term and convolution kernel, so as to dynamically balance features based on the degree of breakage (confidence level or soft mask), and the overall calculation is formulated as a classic convolution implementation with an additional attention term to describe local structure. Besides, the proposed CSCA, not only takes advantage of the contextual attention module, but also considers cross-scale information transfer to generate reasonable features for damaged areas, thus alleviating the inefficiency of the long-range modeling capability of convolutional neural networks. Qualitative and quantitative experiments show that our method performs better than state-of-the-arts, producing clearer, more coherent and visually plausible inpainting results. The code can be found at github.com/dengyecode/CANet_image_inpainting.

Image Inpainting with Bilateral Convolution

Due to sensor malfunctions and poor atmospheric conditions, remote sensing images often miss important information/pixels, which affects downstream tasks, therefore requiring reconstruction. Current image reconstruction methods use deep convolutional neural networks to improve inpainting performances as they have a powerful modeling capability. However, deep convolutional networks learn different features with the same group of convolutional kernels, which restricts their ability to handle diverse image corruptions and often results in color discrepancy and blurriness in the recovered images. To mitigate this problem, in this paper, we propose an operator called Bilateral Convolution (BC) to adaptively preserve and propagate information from known regions to missing data regions. On the basis of vanilla convolution, the BC dynamically propagates more confident features, which weights the input features of a patch according to their spatial location and feature value. Furthermore, to capture different range dependencies, we designed a Multi-range Window Attention (MWA) module, in which the input feature is divided into multiple sizes of non-overlapped patches for several heads, and then these feature patches are processed by the window self-attention. With BC and MWA, we designed a bilateral convolution network for image inpainting. We conducted experiments on remote sensing datasets and several typical image inpainting datasets to verify the effectiveness and generalization of our network. The results show that our network adaptively captures features between known and unknown regions, generates appropriate content for various corrupted images, and has a competitive performance compared with state-of-the-art methods.