摘要近年来,随着电子商务的快速发展,原有的基于文本的商品搜索已经不能够很好的满足大家对更精确寻找商品的需求。如何更加高效的对商品图片进行搜索是一个亟待解决的问题。
本文从这个方面入手,通过使用一般的图像处理算法来进行了商品图像的检索。对基于内容的图像检索系统的系统需求与可行性予以了分析,设计并具体实现了系统的各个功能模块。
系统通过对用户输入的待检索图像使用用户选择的检索方法进行特征提取,利用提取出来的特征在图像库中进行检索,最终返回匹配最接近的显示出来。本系统使用了多种特征提取的方法,包括颜色直方图,形状不变矩,SIFT算子等等。59659
通过实际的实验证明,本系统是可以基本满足基于内容的图像检索的需求。
毕业论文关键词 CBIR,SIFT算子,颜色直方图,形状不变矩
毕业设计说明书(论文)外文摘要
Title The Commodity Images Retrieval Based On Contents
Abstract
As the rapid development of e-commerce and the increasing demand for multi-way of commodity image retrieval online, the traditional commodity image retrieval based on text can no longer bring enough convenience. How to retrieve commodity images has been a problem that needs to be solved.
Aiming to solve the problem, several image retrieval algorithms are proposed. This dissertation did a throughout requirement analysis and feasibility analysis of the image retrieval system, designed and implemented each and every functional module.
This system processes and analyzes the image from the user's input, extracts the features from the source image, which will be used as the search criteria to find the most top ten images shares the similarity. This feature extraction can be done in several different ways, including color histogram, moment invariant, SIFT operator and so on.
Some tests have proved that this system can meet the basic requirements of content based image retrieval.
Keywords:CBIR, SIFT operator, color histogram, moment invariant
1 引言 1
1.1 研究目的 1
1.2 研究背景 1
1.3 研究意义 2
2 相关技术的研究发展状况 2
2.1 图像特征提取技术 3
2.1.1 颜色特征 3
2.1.2 纹理特征 3
2.1.3 形状特征 4
2.1.4 空间关系特性 5
2.1.5 图像特征性能评价 5
2.2 相似度计算技术 6
2.3 SIFT算法 7
3 开发工具和环境 8
3.1 开发工具 8
3.1.1 OpenCV 1.0版本 8
3.1.2 GSL运行库 1.8版本 8
3.2 系统环境的配置 8
3.2.1 OpenCV的配置 8
3.2.2 GSL的配置 10
4 系统总体设计 11
4.1 需求分析