本文選題：圖像匹配 + SURF　； 參考：《安徽理工大學》2017年碩士論文

【摘要】：圖像匹配作為圖像處理技術(shù)的一大分支,近年來在圖像處理領(lǐng)域占據(jù)著重要的地位。許多計算機視覺方面的研究都是在假設(shè)匹配問題已經(jīng)得到解決過后去展開工作的。針對于圖像匹配而言,現(xiàn)今的圖像匹配算法包括兩類:一類是基于灰度的匹配,一類是基于特征點的匹配。本文在圖像匹配方面的研究都是根據(jù)特征點匹配下情況下展開的,對一些經(jīng)典的基于特征點圖像匹配算法作了詳細的研究,然后針對SURF算法進行了改進。首先,本文總結(jié)了國內(nèi)外圖像匹配方面的研究與發(fā)展,并對現(xiàn)下流行的基于特征點的圖像匹配算法分析了國內(nèi)外的研究現(xiàn)狀、常見的研究方法等。其次,重點就一些經(jīng)典的基于特征點的匹配算法的理論原理進行了介紹,主要包括特征點的提取,特征描述符的建立和匹配方法等。同時依賴于計算機視覺開源庫OpenCV開發(fā),通過具體的實驗來對各種匹配算法性能進行對比,分析實驗數(shù)據(jù),評估實驗結(jié)果。最后,依賴于開源視覺庫OpenCV開發(fā),采用算法oFAST檢測特征點與SURF算法建立描述符相結(jié)合,同時在匹配方法中,結(jié)合使用LBPH演化算法,來逐步提高匹配精度。在對改進的SURF算法實驗評估中發(fā)現(xiàn),改進的算法雖然在匹配效率上有所降低,但在匹配精度上有了明顯的提高。
[Abstract]:As a branch of image processing technology, image matching plays an important role in the field of image processing in recent years. Many computer vision studies work on the assumption that the matching problem has been solved. As far as image matching is concerned, there are two kinds of image matching algorithms: one is based on gray level, the other is based on feature points. In this paper, the research on image matching is carried out under the condition of feature point matching. Some classical feature point based image matching algorithms are studied in detail, and then the SURF algorithm is improved. Firstly, this paper summarizes the research and development of image matching at home and abroad, and analyzes the current research situation and common research methods of image matching algorithms based on feature points. Secondly, some classical matching algorithms based on feature points are introduced, including feature point extraction, feature descriptor establishment and matching methods. At the same time, it relies on OpenCV, an open-source computer vision library, to compare the performance of various matching algorithms, analyze the experimental data and evaluate the experimental results through specific experiments. Finally, based on the OpenCV development of open source vision library, the algorithm oFAST is used to detect feature points and SURF algorithm is used to establish descriptors. In the matching method, LBPH evolution algorithm is combined to improve the matching accuracy step by step. In the experiment evaluation of the improved surf algorithm, it is found that although the improved algorithm has lower matching efficiency, the accuracy of the improved algorithm has been improved obviously.
【學位授予單位】：安徽理工大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41

【參考文獻】

相關(guān)期刊論文 前4條

1 彭曉明,丁明躍,周成平,張?zhí)煨?一種利用Hausdorff距離的高效目標搜索算法[J];中國圖象圖形學報;2004年01期

2 余莉,王潤生;基于多尺度變形模板的目標檢測與識別[J];計算機研究與發(fā)展;2002年10期

3 田原,梁德群,吳更石;基于點集不變性匹配的目標檢測與識別方法[J];計算機學報;1999年02期

4 劉海鷹,黃勝華,彭思龍,洪繼光;基于小波多尺度分析的圖象快速匹配模型[J];中國圖象圖形學報;1998年11期

，

本文編號：2083648