【摘要】：隨著信息社會的發(fā)展，圖像的使用己經(jīng)滲透到社會的各行各業(yè)，日益增多的圖像來源為人們提供了豐富的信息。如何快速地搜索有用的圖像己變得越來越迫切。當前流行的網(wǎng)絡搜索引擎大多基于文本，對于基于內(nèi)容的圖像檢索的研究才剛剛起步。將數(shù)字圖像處理、數(shù)據(jù)庫和圖像檢索技術結合起來，建立高速、便捷的圖像搜索引擎具有重要的理論和應用價值。 基于內(nèi)容的圖像檢索技術包括如下幾個內(nèi)容:圖像底層特征提取、相似度匹配、索引機制。本文圍繞圖像底層特征提取展開了研究，將提取顏色特征、紋理特征、形狀特征的經(jīng)典算法進行了融合和改進，并提出了圖像自動分割算法，脈沖耦合神經(jīng)網(wǎng)絡和形態(tài)學相結合，在圖像對象上的標注。在比較分析現(xiàn)有的圖像底層特征的特點的基礎上,提出了基于顏色主色調和形狀特征的綜合特征檢索方法。通過上述特征提取算法可以方便地提取出圖像多方面的特征，，大大增加了圖像檢索的精度。 本文以Visual Studio2010作為開發(fā)工具，Matlab7.0作為實驗工具，設計了一個簡單的基于內(nèi)容的圖像檢索系統(tǒng)。本文設計的CBIR系統(tǒng)基于上述算法實現(xiàn)了圖像特征提取，通過允許用戶輸入?yún)?shù)的方法支持用戶的個性化搜索。最后對本系統(tǒng)進行了實驗，達到了預期的效果。
[Abstract]:With the development of information society, the use of images has penetrated into various industries of society, and more image sources provide people with a wealth of information. How to search for useful images quickly has become more and more urgent. At present, most of the popular web search engines are based on text, and the research on content-based image retrieval has just started. It is of great theoretical and application value to combine digital image processing, database and image retrieval technology to establish a high speed and convenient image search engine. Content-based image retrieval technology includes the following contents: image bottom feature extraction, similarity matching, index mechanism. In this paper, the extraction of color features, texture features and shape features is studied, and the classical algorithm of extracting color features, texture features and shape features is merged and improved, and an automatic image segmentation algorithm is proposed. Pulse coupled neural network and morphology are combined to mark the image object. On the basis of comparing and analyzing the characteristics of the existing image bottom features, a comprehensive feature retrieval method based on color main tone and shape features is proposed. Through the above feature extraction algorithm, many features of the image can be easily extracted, which greatly increases the accuracy of image retrieval. In this paper, a simple content-based image retrieval system is designed by using Visual Studio2010 as the development tool and Matlab7.0 as the experimental tool. The CBIR system designed in this paper realizes image feature extraction based on the above algorithm, and supports user personalized search by allowing users to input parameters. Finally, the experiment of the system is carried out, and the expected effect is achieved.
【學位授予單位】：湖北民族學院
【學位級別】：碩士
【學位授予年份】：2013
【分類號】：TP391.41

【參考文獻】

相關期刊論文 前4條

1 黃祥林,沈蘭蓀;基于內(nèi)容的圖像檢索技術研究[J];電子學報;2002年07期

2 李向陽,莊越挺,潘云鶴;基于內(nèi)容的圖像檢索技術與系統(tǒng)[J];計算機研究與發(fā)展;2001年03期

3 金韜,任秀麗;圖像檢索中顏色特征的提取與匹配[J];計算機輔助設計與圖形學學報;2000年06期

4 丁險峰,吳洪,張宏江,馬頌德;形狀匹配綜述[J];自動化學報;2001年05期

相關博士學位論文 前3條

1 孫君頂;基于內(nèi)容的圖像檢索技術研究[D];西安電子科技大學;2005年

2 崔江濤;高維索引技術中向量近似方法研究[D];西安電子科技大學;2005年

3 趙珊;基于內(nèi)容的圖像檢索關鍵技術研究[D];西安電子科技大學;2007年

本文編號：2478163