兴趣点检测技术在“以图搜图”系统中的应用研究_毕业论文

毕业论文移动版

毕业论文 > 计算机论文 >

兴趣点检测技术在“以图搜图”系统中的应用研究

摘要随着互联网的发展,简单的文字搜索已经不能满足人们的需求,传统的文字搜索局限于一个或几个关键词,有时无法准确表达人们的需求,故图像搜索是一种必然。
图像由于其包含的信息量大和直观等优势进入人们的搜索视野。但由于编码格式,大小等相差很大,所以很难应用在传统搜索行业。本文通过研究图片搜索算法,实现以图搜图功能,其中由于图像搜索精度始终无法提高,故本文利用兴趣点检测算法应用于“以图搜图”中,具有良好的性能,使图片搜索大大提高了搜索精度。
本文的工作主要有以下几点:1、通过研究和分析图片的颜色、形状和纹理特征,从而将图片的低级信息提取出来。2、深入分析了图片的高级特征,例如SIFT特征,详细介绍了SURF算法,并针对其用于以图搜图系统的提出了优化方案,使用SVM进行分类并进行了相关实验。25866
关键字:兴趣点检测  图像搜索  SIFT  SURF
毕业设计说明书(毕业论文)外文摘要
Title     Points of Interest Detection Technology Applied Research in the "Image Search" system                                               
Abstract
With the development of the Internet, a simple text search can not meet people's needs, the traditional text search limited to one or a few key words, sometimes can not accurately express the people's needs, so the image search is a necessity.
Image because it contains a large amount of information and intuitive advantages into people's search field of vision. Since encoding format and size vary widely, so difficult to apply the traditional search industry. By studying image search algorithm, attempt to search map function, which since the image still can not improve search accuracy, therefore we use interest points detection algorithm is applied to the "attempt to search map", has a good performance, image search industry greatly improve search accuracy.
The main work of this paper are as follows: 1, through the gray and shape features research and analysis of the picture, so that the low-level information is extracted image. 2, in-depth analysis of the advanced features of the picture, such as SURF feature, detailing the SIFT algorithm, and for the attempt to search for the system proposed in Fig optimization, experimental verification.
Keywords: Interest point detection, image search ,SIFT, SURF
目   次
1    绪论    1
1.1 课题研究背景    1
1.2 国内外研究发展情况    3
1.3 研究意义    4
1.4 爱肯栖智慧导购系统的优势    4
1.5  相关技术    5
1.5.1 尺度不变特征转换算法介绍    5
1.5.2 基于颜色特征值的特征提取    5
1.5.3 机器学习    5
1.6 本文的组织结构    6
2.图像分类算法介绍    7
2.1 底层特征分类算法    7
2.1.1 基于颜色特征值的特征提取    7
2.1.2 基于形状特征值的特征提取    7
2.2 相似度的度量    8
2.2.1 Minkowsky 距离    8
2.2.2 直方图相交法    8
2.3  SIFT算法    9
2.3.1 SIFT的特点    9
2.3.2 SIFT的基本思想    9
2.4 SURF算法    13
2.4.1 SURF算法思想    13 (责任编辑:qin)