文档介绍：本科毕业论文
(科研训练、毕业设计)
题目:基于颜色特征的图像检索
方法研究与实现
姓名:
学院:信息科学与技术
系:计算机
专业:计算机
年级:2004
学号:04120074
指导教师(校内): 职称:教授
指导教师(校外) 职称:
年月日
基于颜色特征的图像检索方法研究与实现
【摘要】本文介绍了一个基于改进的加权颜色直方图的图象检索系统。
本系统通过将基于改进的加权的局域颜色直方图的图像检索方法和全局直方图的图像检索方法相结合,提高查全率和查准率。其中,基于分块局域直方图的检索方法利用了图像中间部分的重要性,将图像平均划分成3×3个子块,取中间一块的图像,计算其与参考位图相应位置的颜色特征距离,再计算原图的颜色直方图与参考位图的颜色特征距离,分别赋予权值后得出的值就是图像之间内容的相似程度。本文引入欧氏距离的相似性度量方法实现图像检索。实验表明,该方法具有较好的查全率和查准率。
【关键字】图像检索,颜色特征,颜色直方图,相似性度量方法
CONTENT-BASED IMAGE RETRIEVAL SYSTEM
LEARN AND REALIZATION
ABSTRACT
This image retrieval system is based on a improved weighted color histogram.
bining the improved image retrieval algorithem of based on the weighted central-block color histogram and the image retrieval algorithem of overall histogram, the recall rate and the rate of search is improved . Using the importance of the image of the middle part, the retrieval method based on the block local histogram divide a picture into an average of 3 × 3 sub-blocks. Choose the middle one , calculate the colour feature distance between the middle block of original image and the middle block of reference image .Then calculate the colour feature distance between the original image and reference image. Weighted with fixed value , we can get the similarity between them. We introduce the Euclidean distance measurement methods to achieve similar image retrieval.
The experiments show that the method has a better recall rate and the rate of search.
Keywords: Image Retrieval, features of color,color histogram, similar methods of measurement
目录
摘要--------------------------------------------------------------I
ABSTRACT----------------------------------------------------------II
绪论-------------------------------------------------------1
课题研究的目的和意义---------------------------------------1
图像检索系统的关键技术----------------------------------------
国内外研究现状---------------------------------------------
本文所作的主要工作-----------------------------------------
基于颜色特征的图像检索---