摘要海面粗糙度是表征海洋表面粗糙程度的物理量，主要描述小尺度上海面的起伏情况。目前针对海面粗糙度的研究大多依靠遥感技术实现。

    本文介绍了图像纹理的相关知识，详细阐述纹理分析的几种技术方法，包括结构分析法、模型法、信号处理法和统计法，并通过比较最终选择统计法作为本文的研究方法。

本文选取统计法中的六种分析方法：灰度共生矩阵、灰度梯度方向矩阵、自相关函数、基于FBM亮度差值图像的自相关法、Tamura纹理特征法和基于距离的边缘频率法，并分别选择不同特征参数表征纹理图像粗糙度。最后利用最小二乘支持向量机（LSSVM）做回归拟合，得出相对应的回归模型。69354

    总之，本文从图像监测角度出发，利用几何方法提取特征参数，建立新的模型描述海面粗糙度。

毕业论文关键词：海面粗糙度，统计法，纹理特征，图像粗糙度，LSSVM

毕业设计说明书外文摘要

Title    Image Observation Technology Based on Geometric Analysis                     

Abstract

Sea surface roughness is a physical parameter to represent the roughness of the sea surface, which describes the waves of sea on small scale . At present, the research on sea surface roughness is mostly based on remote sensing technique.

The paper introduces the basic concept of image texture and compare he mainstream techniques of image analysis including : structural analysis , model , signal processing and statistics.statistics is chosen after comparing.

The paper selects six methods of statistics including grey level co-occurrence matrix, gray-gradient direction matrix, sautocorrelation function, method based on FBM’s brightness difference, Tamura texture features andedge frequency method based on the distance,and different characteristic parameters is selected to represent texture and roughness of the image.At last，LSSVM is used doing regression fitting to build corresponding regression model.

In short, geometric methods is used to extract characteristic parameters in this paper,and at the end a new model is built to describe the surface roughness of the sea surface.

Key words: Sea Surface Roughness, Statistical Method, Texture Feature, Image  

           Roughness, LSSVM

目   次

1 引言 1

1.1 研究背景及意义 1

1.2 海面粗糙度 1

1.3 研究现状 2

1.4 论文研究内容 3

2 纹理图像分析技术 5

2.1 纹理定义与分类 5

2.2 纹理特征 5

2.3 纹理分析技术 5

2.4 纹理分析应用领域 10

2.5 本章小结 10

3 图像纹理特征的提取 11

3.1 灰度共生矩阵 11

3.2 灰度梯度方向矩阵 14

3.3 自相关函数 16

3.4 基于分形布朗运动模型的亮度差值图像自相关法