基于边缘信息的结构性纹理合成方法研究

发布时间：2018-04-28 16:03

本文选题：基于边缘信息 + 局部特征　；参考：《山东师范大学》2017年硕士论文

【摘要】：在当今的大数据时代,图像信息非常重要,为此我们需要分析图像,在图像分析中,纹理特征是图像中极为关键的信息,所以纹理分析是图像分析中非常重要的研究分支。由于对于机器视觉和图像处理而言,纹理分析是非常关键的支撑技术,因此对图像纹理分析进行进一步研究不仅理论意义重大,而且应用前景广阔。大量具有相似性的元素互相交织纹理形成了丰富的具有周期性的纹理。究其物理特征,纹理表现为空间位置上像素集合的灰度交替变,反映了像素集合灰度级的空间分布情况,是图像分析的重要指标之一。纹理的表现形式多种多样,作为常见的纹理类型,结构性纹理、随机性纹理和混合型纹理具有各自独特的纹理特性。本文研究主要针对结构性文理,研究了基于边缘信息的结构性纹理合成方法,主要工作如下:1.基于纹理边界的路径追踪算法该方法主要是一种基于纹理边界的路径追踪方法。首先用微分算子求得结构性纹理边缘、提取图像的纹理特征。经过实验并观察结果可证明该方法可行。2.基于边缘信息的结构性纹理合成方法基于边缘信息的结构性纹理合成系统可由以下步骤统领描述:纹理的分类(是否结构性纹理)、纹理特征的提取与计算、特征间的相似性度量、基于边界的纹理合成。其中,纹理特征的提取和计算是其中的关键环节。
[Abstract]:In the era of big data, image information is very important, so we need to analyze images. Texture feature is the most important information in image analysis, so texture analysis is a very important research branch in image analysis. Because texture analysis is a key supporting technology for machine vision and image processing, it is not only of great theoretical significance to further study image texture analysis, but also has broad application prospects. A large number of similar elements interweave with each other to form rich periodic textures. According to its physical characteristics, texture is the alternating change of gray level of pixel set in spatial position, which reflects the spatial distribution of gray level of pixel set, and is one of the important indexes of image analysis. As common texture types, structural texture, random texture and mixed texture have their own unique texture characteristics. In this paper, the structural texture synthesis method based on edge information is studied. The main work is as follows: 1. A path tracking algorithm based on texture boundary this method is mainly based on texture boundary. Firstly, the structural texture edge is obtained by differential operator, and the texture feature of the image is extracted. The experimental results show that the method is feasible. Structural texture Synthesis method based on Edge Information the structural texture Synthesis system based on Edge Information can be described by the following steps: texture classification (whether structural texture is structured texture feature extraction and calculation). Similarity measure between features, texture synthesis based on boundary. Among them, the extraction and calculation of texture feature is the key link.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】