Artificial mosaics

Art often provides valuable insight that can be applied to technological innovations, especially in the fields of image processing and computer graphics. In this paper we present a method to transform a raster input image into a good-quality mosaic: an “artificial mosaic.” The creation of mosaics of artistic quality is challenging because the tiles that compose a mosaic, typically small polygons, must be packed tightly and yet must follow and emphasize orientations chosen by the artist. The proposed method can reproduce the colors of the original image and emphasize relevant boundaries by placing tiles along edge directions. No user intervention is needed to detect the boundaries: they are automatically detected using a simple but effective image processing technique. Several examples reported in the paper show how the right mixture of mathematical tools together with time-tested ideas of mosaicists may lead to impressive results.     

Generating Classic Mosaics with Graph Cuts

Classic mosaic is an old and durable art form. Generating artificial classic mosaics from digital images is an interesting problem that has attracted attention in recent years. Previous approaches to mosaic generation are largely based on heuristics, and therefore it is harder to analyse, predict and improve their performance. In addition, previous methods have a number of disadvantages, such as requiring that the number of tiles in a mosaic is known a priori, or relying on extensive user interaction, or using heuristics for tile placement that lead to visible artefacts. We propose a classic mosaic generation algorithm that is based on a principled global optimization. Our approach is fully automatic. We design and optimize an objective function that incorporates the desired mosaic properties, such as tile alignment to significant image edges, prohibiting tile overlap, etc. Our optimization method is based on graph cuts, which proved to be a powerful optimization tool in graphics and computer vision. Experimental comparison to previous work demonstrate the advantages of our approach.     

多摄像机图像拼接自动色彩均衡算法

针对多摄像机视频图像,分析了图像拼接技术中存在的问题,讨论了目前图像处理中常用的图像色彩均衡算法,提出了一种基于图像像素均值统计的亮度和色彩均衡处理算法。首先提取相邻两摄像机同步帧图像的重叠区域并对重叠区图像进行通道分离（RGB）,把其中一幅作为参考图像,另一幅作为目标图像,分别统计两幅图像各颜色通道像素均值差,用差值强制修正整幅目标图像。然后对修正后的图像和参考图像（整幅图像）进行颜色空间转换（RGB到HSV）,再次统计两幅图像亮度通道（V通道）均值差,用差值强制修正整幅目标图像亮度。实验结果证明,该算法能有效校正相邻摄像机图像的亮度和色差,对后期的拼接融合处理起到了很好的改善效果。     

Mesh-guided optimized retexturing for image and video

This paper presents an approach of replacing textures of specified regions in the input image and video using stretch-based mesh optimization.. The retexturing results have the similar distortion and shading effects conforming to the underlying geometry and lighting conditions. For replacing textures in single image,two important steps are developed: the stretch-based mesh parametrization incorporating the recovered normal information is deduced to imitate perspective distortion of the region of interest; the Poisson-based refinement process is exploited to account for texture distortion at fine scale.The luminance of the input image is preserved through color transfer in YCbCr color space. Our approach is independent of the replaced textures. Once the input image is processed, any new texture can be applied to efficiently generate the retexturing results. For video retexturing, we propose key-frame-based texture replacement extended and generalized from the image retexturing. Our approach repeatedly propagates the replacement result of key frame to the rest of the frames. We develop the local motion optimization scheme to deal with the inaccuracies and errors of robust optical flow when tracking moving objects. Visibility shifting and texture drifting are effectively alleviated using graphcut segmentation algorithm and the global optimization to smooth trajectories of the tracked points over temporal domain. Our experimental results showed that the proposed approach can generate visually pleasing results for both image and video.     

Augmented Lagrangian Method for Total Variation Based Image Restoration and Segmentation Over Triangulated Surfaces

Recently total variation (TV) regularization has been proven very successful in image restoration and segmentation. In image restoration, TV based models offer a good edge preservation property. In image segmentation, TV (or vectorial TV) helps to obtain convex formulations of the problems and thus provides global minimizations. Due to these advantages, TV based models have been extended to image restoration and data segmentation on manifolds. However, TV based restoration and segmentation models are difficult to solve, due to the nonlinearity and non-differentiability of the TV term. Inspired by the success of operator splitting and the augmented Lagrangian method (ALM) in 2D planar image processing, we extend the method to TV and vectorial TV based image restoration and segmentation on triangulated surfaces, which are widely used in computer graphics and computer vision. In particular, we will focus on the following problems. First, several Hilbert spaces will be given to describe TV and vectorial TV based variational models in the discrete setting. Second, we present ALM applied to TV and vectorial TV image restoration on mesh surfaces, leading to efficient algorithms for both gray and color image restoration. Third, we discuss ALM for vectorial TV based multi-region image segmentation, which also works for both gray and color images. The proposed method benefits from fast solvers for sparse linear systems and closed form solutions to subproblems. Experiments on both gray and color images demonstrate the efficiency of our algorithms.     

Efficient paper mosaic rendering on mobile devices based on position-based tiling

According to the growth of mobile technology, people experience high-performance mobile devices. In this environment, processing in real-time is required for mobile application. For colored paper mosaic application on mobile devices, the problem regarding speed must be solved while maintaining high-quality. In this paper, we propose an accelerated method that generates tiles based on their position. We locate tiles by considering the edges of the image and the shapes of neighbor tiles. The shape of the tiles is determined based on the position of each tile. In addition, we display the generation process of a result for our method be seemed real-time method. In here, we propose an ordering method that is similar to the ordering method of humans by considering color, edge, distance and direction.     

Real‐time Image‐based Lighting of Microfacet BRDFs with Varying Iridescence

Iridescence is a natural phenomenon that is perceived as gradual color changes, depending on the view and illumination direction. Prominent examples are the colors seen in oil films and soap bubbles. Unfortunately, iridescent effects are particularly difficult to recreate in real‐time computer graphics. We present a high‐quality real‐time method for rendering iridescent effects under image‐based lighting. Previous methods model dielectric thin‐films of varying thickness on top of an arbitrary micro‐facet model with a conducting or dielectric base material, and evaluate the resulting reflectance term, responsible for the iridescent effects, only for a single direction when using real‐time image‐based lighting. This leads to bright halos at grazing angles and over‐saturated colors on rough surfaces, which causes an unnatural appearance that is not observed in ground truth data. We address this problem by taking the distribution of light directions, given by the environment map and surface roughness, into account when evaluating the reflectance term. In particular, our approach prefilters the first and second moments of the light direction, which are used to evaluate a filtered version of the reflectance term. We show that the visual quality of our approach is superior to the ones previously achieved, while having only a small negative impact on performance.     

Image quality evaluation for OLED-based smart-phone displays at various lighting conditions

The image quality of three organic light-emitting diode (OLED) based smart-phone displays was assessed at three levels of ambient lighting conditions corresponding to the darkroom, indoor and outdoor environment, respectively. Seven perceptual attributes, i.e., naturalness, colorfulness, brightness, contrast, sharpness, preference, and overall image quality (IQ), were evaluated in both standard dynamic range (SDR) and high dynamic range (HDR) mode via psychophysical experiments by rank order method, while readability was assessed only in SDR mode and gradation was investigated only in HDR mode. The experimental results demonstrate that, besides the color gamut, the tone reproduction curve is also an important factor affecting the colorfulness of mobile display in the two modes. Higher peak luminance would not mean better performance on brightness and contrast for HDR images, which is opposite to SDR mode. Further analysis of variance (ANOVA) indicates that the ranking results of all perceptual attributes are not significantly affected by the ambient lighting levels in both SDR and HDR modes.     

Discrete Line Congruences for Shading and Lighting

Two‐parameter families of straight lines (line congruences) are implicitly present in graphics and geometry processing in several important ways including lighting and shape analysis. In this paper we make them accessible to optimization and geometric computing, by introducing a general discrete version of congruences based on piecewise‐linear correspondences between triangle meshes. Our applications of congruences are based on the extraction of a so‐called torsion‐free support structure, which is a procedure analogous to remeshing a surface along its principal curvature lines. A particular application of such structures are freeform shading and lighting systems for architecture. We combine interactive design of such systems with global optimization in order to satisfy geometric constraints. In this way we explore a new area where architecture can greatly benefit from graphics.     

基于B样条曲线模型的彩色图像分割

提出一种可靠的色彩建模方法,建立基于亮度变化的色彩统计特征数学模型,其特征包含随亮度变化的色度、饱和度的均值以及标准偏差,应用B样条对模型进行曲线拟合,实现彩色图像分割。实验结果表明,该算法的分割效果较好,鲁棒性强。能够适应不同光照条件下彩色图像的分割。     

基于色彩学习的彩色图象分割方法

图象分割是实现计算机图象识别与理解的基础，而彩色是进行图象分割的一个重要手段。本文给出了一种基色彩学习的彩色图象分割算法。     

基于颜色量化和密度峰聚类的彩色图像分割

彩色图像分割是簇绒地毯数字化制造的关键技术,图像的分割质量直接影响到后续的图像处理。为解决地毯的彩色图像分割问题,针对人眼在RGB颜色空间中感知不均匀的特性,提出了一种基于颜色量化和密度峰聚类的彩色图像分割算法。基于Lab颜色空间进行颜色量化,在HVC颜色空间中用NBS距离来衡量人眼对颜色差异的感知程度,采用改进的密度峰聚类算法自动确定聚类中心,从而分割地毯图案。实验结果表明,该算法能在不影响人眼感知的前提下得到颜色种类少且边缘清晰的地毯分割图像。     

Adapting softcopy color reproduction to ambient illumination

Abstract— The perceived colors of an image seen on a self‐luminous display are affected by ambient illumination. The ambient light reflected from the display faceplate is mixed with the image‐forming light emitted by the display. In addition to this direct physical effect of viewing flare, ambient illumination causes perceptual changes by affecting the adaptation state of the viewer's visual system. This paper first discusses these effects and how they can be compensated, outlining a display system able to adjust its output based on prevailing lighting conditions. The emphasis is on compensating for the perceptual effects of viewing conditions by means of color‐appearance modeling. The effects of varying the degree of chromatic adaptation parameter D and the surround compensation parameters c and N_c of the CIECAM97s color‐appearance model were studied in psychophysical experiments. In these memory‐based paired comparison experiments, the observers judged the appearance of images shown on an LCD under three different ambient‐illumination conditions. The dependence of the optimal parameter values on the level of ambient illumination was evident. The results of the final experiment, using a category scaling technique, showed the benefit of using the color‐appearance model with the optimized parameters in compensating for the perceptual changes caused by varying ambient illumination.     

一种融合聚类与区域生长的彩色图像分割方法

论文提出了一种将聚类和区域生长有机融合的彩色图像分割方法。为了捕获图像的纹理特征,首先将图像划分成16×16子块,然后在块中按照视觉一致性准则进行颜色聚类,对于聚类后的子块,提取其颜色与纹理特征,然后采用符合人类视觉特征的生长规则,进行基于子块的区域生长。该方法充分利用了聚类算法和区域生长算法的各自优点,并符合人类视觉特征的分割策略。利用提出的算法对多幅自然图像进行了分割实验,实验结果证明了算法的有效性。     

Perceptually Motivated BRDF Comparison using Single Image

Surface reflectance of real‐world materials is now widely represented by the bidirectional reflectance distribution function (BRDF) and also by spatially varying representations such as SVBRDF and the bidirectional texture function (BTF). The raw surface reflectance measurements are typically compressed or fitted by analytical models, that always introduce a certain loss of accuracy. For its evaluation we need a distance function between a reference surface reflectance and its approximate version. Although some of the past techniques tried to reflect the perceptual sensitivity of human vision, they have neither optimized illumination and viewing conditions nor surface shape. In this paper, we suggest a new image‐based methodology for comparing different anisotropic BRDFs. We use optimization techniques to generate a novel surface which has extensive coverage of incoming and outgoing light directions, while preserving its features and frequencies that are important for material appearance judgments. A single rendered image of such a surface along with simultaneously optimized lighting and viewing directions leads to the computation of a meaningful BRDF difference, by means of standard image difference predictors. A psychophysical experiments revealed that our surface provides richer information on material properties than the standard surfaces often used in computer graphics, e.g., sphere or blob.     

语义分割和HSV色彩空间引导的低光照图像增强

目的 低光照图像增强是图像处理中的基本任务之一。虽然已经提出了各种方法,但它们往往无法在视觉上产生吸引人的结果,这些图像存在细节不清晰、对比度不高和色彩失真等问题,同时也对后续目标检测、语义分割等任务有不利影响。针对上述问题,提出一种语义分割和HSV（hue,saturation and value）色彩空间引导的低光照图像增强方法。方法 首先提出一个迭代图像增强网络,逐步学习低光照图像与增强图像之间像素级的最佳映射,同时为了在增强过程中保留语义信息,引入一个无监督的语义分割网络并计算语义损失,该网络不需要昂贵的分割注释。为了进一步解决色彩失真问题,在训练时利用HSV色彩空间设计HSV损失;为了解决低光照图像增强中出现细节不清晰的问题,设计了空间一致性损失,使增强图像与对应的低光照图像尽可能细节一致。最终,本文的总损失函数由5个损失函数组成。结果 将本文方法与LIME（low-light image enhancement）、RetinexNet（deep retinex decomposition）、EnlightenGAN（deep light enhancement using generative adversarial networks）、Zero-DCE（zero-reference deep curve estimation）和SGZ（semantic-guided zero-shot learning）5种方法进行了比较。在峰值信噪比（peak signal-to noise ratio,PSNR）上,本文方法平均比Zero-DCE（zero-reference deep curve estimation）提高了0.32dB;在自然图像质量评价（natural image quality evaluation,NIQE）方面,本文方法比EnlightenGAN提高了6%。从主观上看,本文方法具有更好的视觉效果。结论 本文所提出的低光照图像增强方法能有效解决细节不清晰、色彩失真等问题,具有一定的应用价值。     

基于HIS模型的火焰图像分割算法研究

在火灾识别系统中,为了更加方便地进行色彩与轮廓的处理和识别,将HIS模型作为基本的颜色空间,HIS模型是从人的视觉系统出发,用色调、饱和度和亮度来描述色彩的,最后提出了基于HIS模型的火灾图像分割算法。     

基于子块的区域生长的彩色图像分割算法

提出了一种基于图像子块的区域生长算法,应用于彩色图像分割。首先将图像划分成多个不重叠子块,然后利用从CIE L*a*b*颜色空间中提取出的每个子块的颜色和纹理特征,先进行子块内颜色聚类,达到子块分类的目的,再根据生长准则进行基于分类子块的区域生长,实现对自然彩色图像的分割。实验结果证明了算法的有效性,分割结果符合人的主观感知。     

Image segmentation and similarity of color-texture objects

We aim for content-based image retrieval of textured objects in natural scenes under varying illumination and viewing conditions. To achieve this, image retrieval is based on matching feature distributions derived from color invariant gradients. To cope with object cluttering, region-based texture segmentation is applied on the target images prior to the actual image retrieval process. The retrieval scheme is empirically verified on color images taken from textured objects under different lighting conditions.     

数字图像处理系统设计与实现

针对数字图像处理的需求,开发设计了基于MATLAB GUI的数字图像处理系统.系统模块中包含对文件进行编辑,包含对图像大小调整,裁剪,灰度化处理,以及显示图像的边界图,图像进行类型转换,颜色空间转换.对数字图像处理的设计主界面中一共包含以下几个实验项目,图像的变换、增强、分割、其他常用处理.图像的傅里叶变换,离散余弦变...