基于纹理的图像分割方法研究-电路与系统专业论文.docxVIP

万方数据 万方数据 摘要 摘 要 纹理图像分割对于图像处理、模式识别和许多计算机视觉的研究具有重要意 义，一直以来都是人们研究的热点问题。在实际生活中也己经得到了广泛的应用， 其应用包括医学诊断中的癌细胞识别、从遥感图片中识别有用的军事目标和民用 目标等许多领域。 纹理图像分割就是把一幅纹理类别不同的图像划分成几个区域，其中，每一 个区域都具有一致或相似的纹理特性。纹理图像分割的过程一般由两部分组成： 特征提取和区域一致性分割算法。纹理图像分割的结果就是给图像的每个像素分 配一个类别的标记。 撰写本文前依次做了如下工作：研究了迄今为止主要的纹理特征提取方法， 例如空间自相关法、傅里叶功率普法、共生矩阵法、基于邻域特征统计的纹理分 析法、灰度差分统计法、行程长度统计法、分数维描述纹理法、Tamura 纹理特征 法；研究了人类视觉系统（Human Vision System，HVS）的信息处理机制，即基 于 HVS 的“滤波器-全波整流-滤波器”（Filter-Rectify-Filter，FRF）的纹理 分割模型；编写了 MATLAB 程序，实现了基于 FRF 模型的纹理分割算法；利用 Brodatz 的纹理图像库对编写的 MATLAB 程序进行测试，验证了基于 FRF 模型的 纹理分割算法的正确性和分辨率。同时，编写了 VC++程序，实现了基于 FRF 模 型的纹理分割算法，且已成功移植到实验室承担项目的集成验证平台。 关键词：视觉处理机制，Gabor 滤波器，纹理特征，图像分割 I ABSTRACT ABSTRACT Nowadays Texture image segmentation is a hot research topic, because it is important for Image Processing, Pattern Recognition and Computer Vision. There is a lot of application about it, such as cancer cell recognition in medical diagnosis, distinguish military objectives and civilian targets from remote sensing image, etc. Texture image segmentation means that an image which is composed by different textures is divided into several regions which have the same or similar texture. Texture image segmentation contains two parts: feature extraction and region consistency segmentation algorithm. The result of Texture image segmentation is a set which contains labels that are used to label every pixel. Before this paper, the following works have been done: investigated main texture feature extraction methods by now, such as Spatial autocorrelation, Fourier power spectrum, Co-occurrence matrix, Neighborhood characteristics based on statistical texture analysis, Gray level difference statistics, stroke length statistics, fractal dimension texture description method, Tamura texture features method; investigated the processing mechanism of the human visual system; investigated Filter - Full-wave rectifier - filter (FRF) model of texture segmentation based on HVS; written MATLAB program to achieve texture segmentation