基于EBP-OP算法与纹理特征的图像检索

基于纹理特征的图像检索在多媒体数据库管理和多媒体通信传输中得到越来越多的重视。文章提出了一种基于图像纹理特征与多层前馈神经网络EBP-OP算法的图像检索方法。首先运用灰度共生矩阵提取图像的纹理特征,然后用EBP-OP算法对提取的纹理特征进行分类,并在此基础上实现一组纹理图像的检索,实验证明这种方法是有效的。     

基于随机聚类的结肠镜图像分割技术的研究

针对彩色结肠镜图像，提出一种融合颜色、亮度、空间距离和纹理信息，采用随机聚类的彩色图像分割新算法。该算法采用分形维作为图像纹理特征的度量，用一种基于元素间的相似性的随机聚类方法对特征空间进行聚类。该聚类算法是基于对图中的切割进行采样的新的图论算法，可以自动获得最佳的分类数目。     

基于无参数Markov随机场模型的彩色纹理综合方法

提出一种基于多尺度、无参数Markov随机场模型和KL变换的彩色纹理综合方法。该方法能捕捉原始纹理的高阶统计特性,综合出与原始纹理视觉一致的彩色纹理图像。实验结果证明该方法对彩色纹理图像的综合非常有效。     

基于颜色统计不变性的纹理图像光照分析

针对目前纹理光照分析方法计算效率慢且需要过多用户交互的缺点,提出一种基于颜色统计不变性的快速纹理光照估算方法。首先,定义纹理图像的颜色统计特征、颜色统计不变性和窗口稳定尺寸,据此估算白色光照和彩色光照图像;其次,通过对纹理图像上-下采样的方法,降低彩色光照估算和本征图像提取的计算复杂度。实验结果表明该方法不但缓解了光照图像中含有纹理结构信息的问题,同时有效提高了运算效率。     

基于四元数Gabor滤波的彩色纹理特征提取

现有的彩色图像纹理特征提取方法是将彩色图像转换为灰度图像或者对彩色图像进行分通道处理,这样的处理方法会丢失原图像的颜色信息和各通道间的相关性,导致特征图像的纹理特征和原图像的纹理特征差异较大。基于上述问题,提出了一种四元数Gabor彩色纹理特征提取方法。首先,根据Gabor滤波和四元数欧拉公式,推导出四元数Gabor滤波,并将彩色图像用四元数矩阵表达;其次提出四元数Gabor滤波卷积算法处理彩色图像,得到多尺度多方向的彩色纹理特征图像;最后对得到的彩色纹理特征图像进行Tamura统计特征的提取。实验结果表明,该方法可以很大程度地保留原图像的粗糙度、对比度和方向度等纹理特征,同时可以提取到原图像的颜色信息。在转化为灰度图像后,该方法在保留粗糙度、对比度和方向度等纹理特征方面优于传统Gabor方法和LBP方法。     

基于多尺度结构张量的多类无监督彩色纹理图像分割方法

由于自然图像中包含丰富的颜色信息与多尺度的纹理信息,伴随多个同质目标区域的出现,依靠半监督手动交互标记的图像分割方法难以实现自动分割,因此提出一种多类无监督彩色纹理图像分割方法.首先,提取了颜色特征(向量结构)与多尺度纹理特征(矩阵集合),并对两者分别进行能量描述;对于2种具有不同特征结构的能量函数,通过多类融合策略计算两者的融合因子,并自适应地融合;再将融合后能量函数的全局最优化问题转化为其对应的多层Graph Cut图割模型,利用最大流?最小割理论计算得到全局近似最优解.为了自适应地控制分割过程,提出了自适应迭代分割收敛准则,并最终在自然图像及合成的彩色纹理图像上进行了质量评估与量化分析.实验结果表明,该方法具有较好的目标区域完整性与一致性,并具有较高的准确率.     

Gabor滤波器在彩色纹理表面缺陷检测中的应用

提出了彩色纹理表面自动缺陷检测的Gabor滤波方法,以同时测度图像中的颜色和纹理偏差.提出的方法不依赖于纹理特征的提取,它基于Gabor滤波器和两个颜色特征复数的彩色图像卷积的能量响应,将彩色纹理图像缺陷检测的复杂问题转换成了滤波图像中的简单的二值化问题.对纺织品、木材等许多实际彩色纹理表面的实验结果表明了该方法的有效性.     

基于Gabor滤波的灰度图像彩色化方法

提出一种基于Gabor滤波的分层次图像彩色化方法.使用Gabor滤波提取图像纹理特征,利用纹理特征将目标灰度图像分割为多个子区域.其中,每个子区域与源参考图像的单个纹理区域相对应,且拥有单调的色彩;分别对子区域进行彩色化,从而降低全局图像彩色化的复杂度.实验结果表明,此方法能抑制图像颜色传输时颜色的丢失,有效地提高彩色化质量.     

多传感器彩色纹理色彩差异修正方法

针对三维彩色扫描仪多个传感器获得的彩色纹理图像存在色彩差异,导致重建的三维彩色模型表面出现色彩分块现象的问题,提出一种基于色彩迁移的纹理图像色彩修正方法。首先,利用综合质量图像评价函数(CAF)进行图像质量评价,从多个传感器获得的彩色纹理图像中选出一幅质量最佳的作为标准图像;然后,参照该图像调整其他纹理图像各个色彩通道的色彩均值、方差等统计量。将所提方法用于三维人体彩色扫描仪的纹理图像色彩修正,结果表明,色彩修正后纹理图像间的色彩差异减小,得到的三维人体彩色模型色彩分块现象得到明显改善,色彩更加均衡自然。与经典色彩变换方法、改进色彩变换方法、基于最小角度选择的融合方法相比,实验的主观效果和客观图像综合质量评价值都表明了该方法的优越性。     

彩色扫描文档图像中图文分割算法

针对彩色扫描文档图像的背景噪声以及文本区的纹理干扰，提出一种利用图像处理技术、结合彩色文档图像自身特点、对文档插图区域进行定位与分割的方法．首先生成一组减弱了文本区纹理信息的多尺度特征缩图；然后采用基于连通度的标记分割法去掉文本区域信息，确定图像区域；最后融合多尺度缩图信息实施图文分割．实验结果表明：该方法对于提高扫描文档图像的压缩比是行之有效的．     

New methods for analysing colour texture based on the Karhunen-Loeve transform and quantification

In this article, we offer an original study on the analysis of the texture of colour images based on Local Linear Transforms (LLT). Our colour approach is based on the separability of the data which reduces the number of texture parameters. We also propose the extension of Run Lengths (RL) and Co-occurrence Matrixes (CM) to colour images. In this respect, two different ways were explored (data merging and quantification). We finally present a comparative study showing the efficiency of the first method (LLT) as well as the complementary nature of the other methods (RL, CM).     

A new benchmark image test suite for evaluating colour texture classification schemes

Several image test suites are available in the literature to evaluate the performance of classification schemes. In the framework of colour texture classification, OuTex-TC-00013 (OuTex) and Contrib-TC-00006 (VisTex) are often used. These colour texture image sets have allowed the accuracies reached by many classification schemes to be compared. However, by analysing the classification results obtained with these two sets of colour texture images, we have noticed that the use of colour histogram yields a higher rate of well-classified images compared to colour texture features. It does not take into account any texture information in the image, this incoherence leads us to question the relevance of these two benchmark colour texture sets for measuring the performances of colour texture classification algorithms. Indeed, the partitioning used to build these two sets consists of extracting training and validating sub-images of an original image. We show that such partitioning leads to biased classification results when it is combined with a classifier such as the nearest neighbour. In this paper a new relevant image test suite is proposed for evaluating colour texture classification schemes. The training and the validating sub-images come from different original images in order to ensure that the correlation of the colour texture images is minimized.     

A learning-based colour image segmentation with extended and compact structural tensor feature representation

In this paper a novel Tensor-Based Image Segmentation Algorithm (TBISA) is presented, which is dedicated for segmentation of colour images. A purpose of TBISA is to distinguish specific objects based on their characteristics, i.e. shape, colour, texture, or a mixture of these features. All of those information are available in colour channel data. Nonetheless, performing image analysis on the pixel level using RGB values, does not allow to access information on texture which is hidden in relation between neighbouring pixels. Therefore, to take full advantage of all available information, we propose to incorporate the Structural Tensors as a feature extraction method. It forms enriched feature set which, apart from colour and intensity, conveys also information of texture. This set is next processed by different classification algorithms for image segmentation. Quality of TBISA is evaluated in a series of experiments carried on benchmark images. Obtained results prove that the proposed method allows accurate and fast image segmentation.     

敏感图像信息识别模型研究

为减少和杜绝色情图像在公共网络上传播的机会,在分析色情图像特征基础上,设计了一个敏感图像信息识别模型,综合使用肤色模型匹配和纹理判别技术,提取多个统计特征,采用SVM进行分类实现了对敏感图像的识别.该模型能够有效识别网络色情图像,适合网络环境下使用,为网络敏感信息监控提供了比较完整的图像识别解决方案.     

Induction operators for a computational colour–texture representation

The aim of this paper is to outline a perceptual approach to a computational colour–texture representation based on some colour induction phenomena. The extension of classical grey level methods for texture processing to the RGB channels of the corresponding colour texture is not the best solution to simulate human perception. Chromatic induction mechanisms of the human visual system, that has been widely studied in psychophysics, play an important role when looking at scenes where the spatial frequency is high as it occurs on texture images. Besides others, chromatic induction includes two complementary effects: chromatic assimilation and chromatic contrast. While the former has been measured by Wandell and Zhang [A spatial extension of CIELAB for digital colour image reproduction, in: SID, 1996] and extended to computer vision by Petrou et al. [Perceptual smoothing and segmentation of colour textures, in: 5th European Conference on Computer Vision, Freiburg, Germany, 1998, pp. 623] as a perceptual blurring, some aspects on the last one still remain to be measured, but it has to be a computational operator that simulates the contrast induction phenomenon performing a perceptual sharpening that preserves the structural properties of the texture. Applying both, the perceptual sharpening and the perceptual blurring, we propose to build a tower of images as an induction front-end that can be the basis of a perceptual representation of colour–textures.     

A region-based image database system using colour and texture

In this paper, a region-based image database system is presented. When an image is entered into the database it is first clustered into similar looking regions using the local colour and texture properties of the image. The mean colour and texture properties of these regions along with their location and size are then stored as an index into that image.To query the database, the user specifies an object on which the same texture and colour properties are calculated. A neural network is trained on these features and then used to search through the indices of the database images. All similar looking regions in the images are shown as the results to the query.     

Combining spatial and colour information for content based image retrieval

Colour is one of the most important features in content based image retrieval. However, colour is rarely used as a feature that codes local spatial information, except for colour texture. This paper presents an approach to represent spatial colour distributions using local principal component analysis (PCA). The representation is based on image windows which are selected by two complementary data driven attentive mechanisms: a symmetry based saliency map and an edge and corner detector. The eigenvectors obtained from local PCA of the selected windows form colour patterns that capture both low and high spatial frequencies, so they are well suited for shape as well as texture representation. Projections of the windows selected from the image database to the local PCs serve as a compact representation for the search database. Queries are formulated by specifying windows within query images. System feedback makes both the search process and the results comprehensible for the user.