本科毕业设计论文 基于内容的图像检索系统及sift算法的应用.docVIP

基于SIFT算法的图像检索系统的研究 摘 要 人类进入现代社会以来，科技、经济、文明不断发展，信息量的增长速度也随之加快。从海量数据中迅速检索出有用信息成为人们的迫切需求。因此，图像检索作为一种能从海量图像中迅速查找到需要图像的技术很快成为人们研究的热点。 在一个图像检索系统中最重要的环节便是图像特征提取与匹配，而SIFT算法正是一种由David Lowe提出的图像局部特征描述子，因为其良好的独特性和鲁棒性而受到人们的普遍关注。本文给出了一种基于SIFT算法的图像检索技术，并依此开发了一套图像检索实验系统，主要研究工作总结为：首先分析了CBIR系统的组成和关键技术的基本原理，重点分析了特征提取环节的技术细节，对比分析了几种经典的特征描述算子，尤其是SIFT算法的基本原理和各自的优缺点。然后根据对比结果选择SIFT算法作为最终的特征提取方案，最后实现了一个实验性的CBIR系统。 实验结果表明SIFT 算法对图像的平移、旋转、视角等变化具有良好的鲁棒性，提出的基于SIFT算法的CBIR系统方案具有良好的效果。 关键词 图像检索；CBIR；SIFT算法；特征提取 Research about Image Retrieval System Based on SIFT Algorithm Abstract Since the humans enter the modern age,the growth of science,technology, economy and civilization has accelerated the growth of information.Retrieving useful information from the huge amounts of data quickly becomes peoples urgent needs.Therefore, image retrieval as a technology which able to find the required image from massive image quickly has became a research focus soon. Image feature extraction and matching is the most important principle in image retrieval system,and SIFT algorithm is a local characteristic of image description made by David Lowe,because of its unique and robustness,SIFT algorithm has admired widespread concern.This article will present an image retrieval technology based on SIFT algorithm and so on develops a testing system for image retrieval.Research summary:Firstly,analysis the basic principle and key technology of CBIR system,focusing on aspects of feature extraction from technical details.Secondly,comparison and analysis several classic describes of the characteristics,particularly foundations and their respective advantages and disadvantages,then select SIFT algorithm based on comparison results as the final programme of feature extraction.Finally achieved an experimental CBIR system. Experimental results show that the SIFT algorithm has good robustness to image translation, rotation and perspective changes,and the CBIR system based on SIFT algorithm also shows good property and action