A fuzzy approach to digital image warping

Digital image warping addresses the problem of how to smoothly transform one digital image into another. The warping process has wide applications in computer animation and can be divided into two classes depending on the type of images being transformed. Gray image warping considers the transformation of one gray-scale image into another-a process also known as image metamorphosis. Binary image warping, on the other hand, addresses the transformation of binary images such as polygonal shapes. The focus in this article is on warping polygons. We can approach the warping of polygonal shapes in two steps. The first establishes a correspondence between the vertices of two given polygons. The second step interpolates the corresponding vertices to generate vertices of an intermediate polygon. This article presents new approaches to both steps. This new algorithm uses fuzzy techniques to warp polygons that have different locations, orientations, sizes and numbers of vertices. The algorithm is robust and extensible to curved shapes     

Artistic Low Poly rendering for images

This paper presents an automatic approach for generating low poly rendering of images, which is particularly popular in the recent art design community. Distinguishing from the traditional image triangulation methods for the sake of compression or vectorization, we propose some critical principles of such Low Poly rendering problem, and simulate the artists creation procedures straightforwardly. To produce the visual effects with clear boundaries, we constrain the vertices along the feature edges extracted from the input image. By employing the Voronoi diagram iteration guided by a feature flow field, the vertices in the result image well reflect the feature structure of the local shape. Moreover, with the salient region detection, we can achieve different mesh densities between the front object and the background. Some special color processing techniques are employed to make our result more artistic. Our method works well on a wide variety of images, no matter raster photographs or artificial images. Experiments show that our approach is able to generate satisfying results similar to the artwork created by professional artists.     

Polygonal representation of digital planar curves through dominant point detection—a nonparametric algorithm

We describe a new algorithm that detects a set of feature points on the boundary of an 8-connected shape that constitute the vertices of a polygonal approximation of the shape itself. The set of feature points (nodes) is a ranked subset of the original shape points whose connected left and right arm extents cover the entire shape. Nodes are ranked based on their strength (in terms of their importance to other boundary points), length of support region, and distance from the centroid. The polygon obtained by linking the detected nodes approximates the contour in an intuitive way. The proposed algorithm does not require an input parameter and works well for shapes with features of multiple sizes.     

基于网络图像的阿奇姆博多拼图

拼图是使用不同形式的素材进行组装的合成艺术.文中介绍了一种崭新的拼图创作方法,通过挑选主题相关的网络图像的前景抠图来表现一幅输入图像,能够生成奇异的具有阿奇姆博多风格的拼图.鉴于网络图像数据的海量规模,总能够找到合适的前景抠图来匹配输入图像的各个分割部分;通过将挑选的前景抠图按照一定的规则进行组装,使其在整体上可以从形状和颜色表现输入图像,而在局部每个前景抠图仍然可以辨识.实验结果和用户体验表明,采用文中方法可以有效地生成具有娱乐性的阿奇姆博多拼图.     

Quantization of color images for display/printing on limited color output devices

A large number of output devices in use today are either bilevel or can produce only a limited number of display levels (gray-scale or color). Most color graphics terminals conforming to Enhanced Graphics Adapter (EGA), Professional Graphics Adapter (PGA), or Video Graphics Array (VGA) standards can display from 16–256 colors, whereas real-world (externally acquired) images constitute typically 16M colors. In this paper, a new color quantization algorithm has been proposed which maps an original image into an output image with a limited number of colors, while still preserving the image quality. The algorithm itself is based on the concepts of vector quantization where a color vector is defined by red, green, and blue components and, based on a random sampling of the input image, a color mapping table is generated. The random sampling provides an estimate of the color distribution of the input image, which is then further combined by a clustering technique to derive the desired number of output colors. A mapping process results in a limited-color output image which is optionally preprocessed (in cases where the number of output colors is very small) by a pseudo-random dithering algorithm rendering a high-quality output. This postprocessing step is particularly useful in images with very few output colors, e.g., 16. Through examples, it is shown that input images with over 16M colors can be easily displayed in as few as 16 colors, with negligible degradation in quality.     

从电场角度引出的基于边缘特征的归一化形状识别算法研究

为了从一新的角度来解决图像形状识别问题，提出了一种从电场角度引出的基于边缘特征归一化的形状识别算法，即在图像处理过程中，像素将被赋予电量，各个图像的形状信息是用带电像素在三维空间产生的电势与电场强度的值来表征。该算法处理都是基于二值图像，并先通过边缘检测技术找出图像边缘信息，然后通过对边缘进行多边形逼近来得出边缘的角点信息，最后再对归一化的观测点进行电势与电场强度的计算。由于一个图形存在着无穷的观测点，因此在实际运用中合理选择观测点是重要且必要的，因为它能更加合理地反映图形的形状信息，故更加适合计算机对图形形状的判断。实验表明，当被识别的二值图像发生旋转、位移、变形时，该方法都能取得比较好的识别效果。     

基于渲染图像角度结构特征的三维模型检索方法

为了充分利用三维模型的颜色、形状、纹理等特征,提出以三维模型渲染图像为数据集,利用渲染图像角度结构特征实现三维模型检索。首先,该方法以三维模型渲染图像为测试集,利用已有类别标记的自然图像作为训练集,通过骨架形状上下文特征对渲染图像进行分类,提取角度结构特征,建立特征库;然后,对输入的自然图像提取角度结构特征,与特征库中的角度结构特征进行相似度匹配计算,实现三维模型检索。实验结果表明, 充分利用 渲染图像的颜色、形状和空间信息是实现三维模型检索的有效方法。     

基于颜色和纹理分析的车牌定位方法

针对复杂背景的车牌定位问题，提出了一种颜色和纹理分析相结合的车牌定位算法。该算法采用基于适合彩色图象相似性比较的HSV颜色模型，首先在颜色空间进行距离和相似度计算；然后对输入图象进行颜色分割，只有满足车牌颜色特性的区域，才进入下一步的处理；最后再利用纹理及结构特征对分割出的颜色区域进行分析和进一步判断，并确定车牌区域。该方法不同于大多数的车牌定位方法，它不仅对车牌的大小、汽车在图象中的位置以及图象背景的限制较少，而且，综合特征定位要比单一特征定位更符合人的视觉要求，因而定位效果更好，应用范围更广。     

基于色调优化的图像视频细节增强

目的 针对已有的细节增强方法难以保持输入图像帧的色调分布的缺点,提出一种基于色调优化的图像视频细节增强算法。方法 首先,为了避免颜色通道的相关性所带来的偏色现象并提高算法效率,对输入图像帧进行颜色空间的转换,提取亮度信息。然后,采用基于局部极值的边缘保持图像滤波方法,快速地将亮度通道图像分解成一幅含有大尺度边缘信息的基图像和多幅含有小尺度细节信息的细节层图像。接着,在用户期望的细节增强系数和输入图像的颜色场的约束下,提出基于梯度域上能量优化的细节增强算法,获得色调一致的细节增强亮度图像。最后,通过颜色空间的逆转换得到最终的细节凸显效果。结果 实验结果表明,本文算法不但能够显著地增强输入图像帧的细节内容,而且能够有效地保持其原有的色调分布,显得更加真实生动。结论 本文算法基本满足科学观察、视频监控和数字视觉特效等领域的技术要求,具有很大的应用潜力。     

一种基于纹理域神经网络的彩色卫星图像分析方法

描述了一种利用纹理域神经网络,通过监督的学习方法来调整和权重,将预处理后成的彩色卫星图像中的纹理信息块进行分类,从而达到在彩色卫星图像中将有用的色彩块分析出来的目的,给出了纹理域的预处理过程,阐述了对原彩色图像进行颜色映射和量化的目标和方法;讨论了纹理域神经网络的构造方法及算法;最后给出了用以上方法对彩色卫星图像进行分类的实验结果,实验证明该方法对彩色卫星图像有较好的分类作用。